fix(withdraw): allow multiple active applications

alembic/versions/ad_reward_boost_round_id.py

@@ -1,47 +0,0 @@
-"""ad_reward_record.boost_round_id(金币膨胀「本轮累计」标签)
-
-看完一条激励视频后客户端要弹「本轮累计获得 N 金币」,N 必须等于这一轮实际到账之和(否则用户
-会认为少发了钱)。单条到账额 reward-result 已经能给,但「一轮」的边界只有客户端知道(点「放弃
-赚钱」才算结束),客户端自己累加又会在进程被杀后丢失。
-
-解法:客户端把轮次 id 随 mediaExtra 透传,穿山甲 S2S 原样带回,发奖时打在记录上;
-reward-result 按 (user_id, boost_round_id) 对 granted 记录求和返回 round_coin。
-
-本列是**纯标签**:不参与发奖判定,发多少/发不发完全不受影响。客户端就算一直复用同一个 id,
-也只是把展示数字滚大,不产生任何新入账(求和的是已发生的发奖记录),无资损风险。
-
-Revision ID: ad_reward_boost_round_id
-Revises: comparison_llm_cost
-Create Date: 2026-07-20
-"""
-from collections.abc import Sequence
-
-import sqlalchemy as sa
-
-from alembic import op
-
-revision: str = "ad_reward_boost_round_id"
-down_revision: str | Sequence[str] | None = "comparison_llm_cost"
-branch_labels: str | Sequence[str] | None = None
-depends_on: str | Sequence[str] | None = None
-
-
-def upgrade() -> None:
-    # 可空、无默认:SQLite 原生支持 ADD COLUMN,不需要 batch_alter_table(同 comparison_llm_cost)。
-    # 存量行留 NULL = 「不属于任何一轮」,求和时天然不参与,老客户端行为不变。
-    op.add_column(
-        "ad_reward_record",
-        sa.Column("boost_round_id", sa.String(length=64), nullable=True),
-    )
-    # 求和恒带 user_id(轮 id 是客户端生成的,不能跨用户信任),故建复合索引而非单列
-    op.create_index(
-        "ix_ad_reward_user_boost_round",
-        "ad_reward_record",
-        ["user_id", "boost_round_id"],
-        unique=False,
-    )
-
-
-def downgrade() -> None:
-    op.drop_index("ix_ad_reward_user_boost_round", table_name="ad_reward_record")
-    op.drop_column("ad_reward_record", "boost_round_id")

alembic/versions/comparison_llm_cost.py

@@ -1,33 +0,0 @@
-"""comparison_record: llm_cost_yuan + llm_price_snapshot(比价 LLM 调用成本 + 当时单价快照)
-
-回填 llm_calls 时按「当时的价」逐模型算出本次比价 LLM 总成本(元),连同所用单价快照一起冻结到
-记录上;admin 比价记录详情展示实际成本(旧记录 NULL → 前端回退估算)。见 services/llm_cost.py。
-
-Revision ID: comparison_llm_cost
-Revises: ad_ecpm_trace_id
-Create Date: 2026-07-13
-"""
-from collections.abc import Sequence
-
-import sqlalchemy as sa
-from sqlalchemy.dialects import postgresql
-
-from alembic import op
-
-revision: str = "comparison_llm_cost"
-down_revision: str | Sequence[str] | None = "ad_ecpm_trace_id"
-branch_labels: str | Sequence[str] | None = None
-depends_on: str | Sequence[str] | None = None
-
-_JSONB = sa.JSON().with_variant(postgresql.JSONB(), "postgresql")
-
-
-def upgrade() -> None:
-    # 均可空、无索引;SQLite 原生支持 ADD COLUMN,无需 batch_alter_table(同 comparison_debug_fields)。
-    op.add_column("comparison_record", sa.Column("llm_cost_yuan", sa.Float(), nullable=True))
-    op.add_column("comparison_record", sa.Column("llm_price_snapshot", _JSONB, nullable=True))
-
-
-def downgrade() -> None:
-    op.drop_column("comparison_record", "llm_price_snapshot")
-    op.drop_column("comparison_record", "llm_cost_yuan")

alembic/versions/drop_signin_boost_record.py

@@ -1,56 +0,0 @@
-"""下线签到膨胀:drop signin_boost_record
-
-产品 2026-07 确认「固定 3000 金币的签到膨胀」从来不是设计内的口径 —— 奖励只有「签到」和
-「看视频」两种。签到弹窗里的「看广告膨胀」改与福利页看视频走同一条 reward_video 路径
-(按 eCPM 公式发,记在 ad_reward_record),signin_boost 场景整体摘除。
-
-⚠️ **只 drop 这张表,不动 coin_transaction**:`biz_type='signin_boost'` 的金币流水是真发过的
-钱,账必须留得住(admin 大盘的 signin_boost_coin_total / signin_boost_watch_count 改为从
-coin_transaction 统计,继续能查回历史)。本表只是「哪天膨胀过」的业务留痕,金额与去向都能
-从流水还原,drop 掉不影响对账。
-
-downgrade 只重建空表结构,**不恢复数据** —— 真要回滚得先从备份捞行。
-
-Revision ID: drop_signin_boost_record
-Revises: ad_reward_boost_round_id
-Create Date: 2026-07-20
-"""
-from collections.abc import Sequence
-
-import sqlalchemy as sa
-
-from alembic import op
-
-revision: str = "drop_signin_boost_record"
-down_revision: str | Sequence[str] | None = "ad_reward_boost_round_id"
-branch_labels: str | Sequence[str] | None = None
-depends_on: str | Sequence[str] | None = None
-
-
-def upgrade() -> None:
-    with op.batch_alter_table("signin_boost_record", schema=None) as batch_op:
-        batch_op.drop_index(batch_op.f("ix_signin_boost_record_user_id"))
-    op.drop_table("signin_boost_record")
-
-
-def downgrade() -> None:
-    # 只还结构不还数据(见模块 docstring)
-    op.create_table(
-        "signin_boost_record",
-        sa.Column("id", sa.Integer(), nullable=False),
-        sa.Column("user_id", sa.Integer(), nullable=False),
-        sa.Column("signin_date", sa.Date(), nullable=False),
-        sa.Column("coin_awarded", sa.Integer(), nullable=False),
-        sa.Column("ad_ref_id", sa.String(length=64), nullable=True),
-        sa.Column(
-            "created_at", sa.DateTime(timezone=True),
-            server_default=sa.text("(CURRENT_TIMESTAMP)"), nullable=False,
-        ),
-        sa.ForeignKeyConstraint(["user_id"], ["user.id"]),
-        sa.PrimaryKeyConstraint("id"),
-        sa.UniqueConstraint("user_id", "signin_date", name="uq_signin_boost_user_date"),
-    )
-    with op.batch_alter_table("signin_boost_record", schema=None) as batch_op:
-        batch_op.create_index(
-            batch_op.f("ix_signin_boost_record_user_id"), ["user_id"], unique=False
-        )

alembic/versions/withdraw_allow_multiple_active.py

@@ -0,0 +1,32 @@
+"""allow multiple active withdraw orders per user
+
+Revision ID: withdraw_allow_multiple_active
+Revises: ad_ecpm_trace_id
+Create Date: 2026-07-12 18:00:00.000000
+
+"""
+from collections.abc import Sequence
+
+import sqlalchemy as sa
+
+from alembic import op
+
+revision: str = "withdraw_allow_multiple_active"
+down_revision: str | Sequence[str] | None = "ad_ecpm_trace_id"
+branch_labels: str | Sequence[str] | None = None
+depends_on: str | Sequence[str] | None = None
+
+
+def upgrade() -> None:
+    op.drop_index("ux_withdraw_order_user_active", table_name="withdraw_order")
+
+
+def downgrade() -> None:
+    op.create_index(
+        "ux_withdraw_order_user_active",
+        "withdraw_order",
+        ["user_id"],
+        unique=True,
+        sqlite_where=sa.text("status IN ('reviewing', 'pending')"),
+        postgresql_where=sa.text("status IN ('reviewing', 'pending')"),
+    )

app/admin/repositories/stats.py

@@ -25,7 +25,7 @@ from app.models.coupon_state import (
 from app.models.cps_order import CpsOrder
 from app.models.feedback import Feedback
 from app.models.savings import SavingsRecord
-from app.models.signin import SigninRecord
+from app.models.signin import SigninBoostRecord, SigninRecord
 from app.models.user import User
 from app.models.wallet import CoinTransaction, WithdrawOrder
 
@@ -590,14 +590,7 @@ def dashboard_overview(
                 CoinTransaction.amount > 0,
                 CoinTransaction.biz_type == "signin_boost",
             ),
-            # 签到膨胀 2026-07 已下线,signin_boost_record 表随之 drop。这两项保留为**历史口径**
-            # (钱是真发过的,账要能查回)。次数改数金币流水:一次膨胀 = 一笔 signin_boost 流水,
-            # 与原来数 signin_boost_record 行数等价。
-            "signin_boost_watch_count": _count(
-                CoinTransaction,
-                CoinTransaction.biz_type == "signin_boost",
-                CoinTransaction.amount > 0,
-            ),
+            "signin_boost_watch_count": _count(SigninBoostRecord),
         },
         "cash": {
             "withdraw_success_cents": _sum(

app/admin/schemas/comparison.py

@@ -36,8 +36,6 @@ class AdminComparisonListItem(BaseModel):
     retry_count: int | None = None
     input_tokens: int | None = None      # Σ usage.prompt_tokens(server 派生)
     output_tokens: int | None = None     # Σ usage.completion_tokens(server 派生)
-    # 本次比价 LLM 总成本(元,按当时价冻结);旧记录/未回填为 None → 前端「成本」列回退估算。见 services/llm_cost.py。
-    llm_cost_yuan: float | None = None
     device_model: str | None = None
     rom_vendor: str | None = None
     rom_name: str | None = None
@@ -74,5 +72,3 @@ class AdminComparisonDetail(AdminComparisonListItem):
     # 原始上报全量;「卡在哪一步」从 raw_payload.platform_results[*].status 读
     # (store_not_found/items_not_found/below_minimum/unsupported = 卡在 找店/加菜/起送/读价)。
     raw_payload: dict | None = None
-    # 算成本所用单价快照 {mode, prices:{model:{...}}}(llm_cost_yuan 继承自列表项)。见 services/llm_cost.py。
-    llm_price_snapshot: dict | None = None

app/api/v1/ad.py

@@ -3,8 +3,6 @@
 路由前缀 `/api/v1/ad`:
   GET  /pangle-callback   穿山甲 S2S 发奖回调(**无 JWT,靠验签**),穿山甲服务器调
   GET  /reward-status     客户端查今日看广告发奖进度(Bearer)
-  GET  /reward-result/{ad_session_id}
-                          客户端按会话查本次广告实发金币(Bearer,只读,弹窗金额用)
 
 发奖走服务端:激励视频播完穿山甲回调本接口,验签通过后幂等发金币。客户端只负责
 看完后刷新余额,不参与发奖,被破解也刷不到钱。
@@ -15,7 +13,7 @@ import json
 import logging
 import uuid
 
-from fastapi import APIRouter, Depends, HTTPException, Path, Request, status
+from fastapi import APIRouter, Depends, HTTPException, Request, status
 
 from app.api.deps import CurrentUser, DbSession
 from app.core import rewards
@@ -27,8 +25,8 @@ from app.repositories import ad_feed_reward as crud_feed
 from app.repositories import ad_reward as crud_ad
 from app.repositories import ad_watch as crud_watch
 from app.repositories import app_config
+from app.repositories import signin as crud_signin
 from app.schemas.ad import (
-    AdRewardResultOut,
     AdRewardStatusOut,
     EcpmReportIn,
     EcpmReportOut,
@@ -54,14 +52,11 @@ REASON_BAD_PARAMS = 1  # 验签过但缺 trans_id / user_id 非数字
 REASON_UNKNOWN_USER = 2  # user_id 不存在(可能伪造)
 
 REWARD_SCENE_REWARD_VIDEO = "reward_video"
+REWARD_SCENE_SIGNIN_BOOST = "signin_boost"
 # 提现看视频:看完才能提现的「硬门槛」广告,**不发金币**,只记一条幂等记录(收益由 eCPM 上报口径
 # ad_type="withdrawal_video" 单独统计)。故意不放进 SUPPORTED_REWARD_SCENES——它不走发币分支。
 REWARD_SCENE_WITHDRAWAL_AD = "withdrawal_ad"
-# 2026-07 下线 signin_boost(签到膨胀):它按固定 3000 金币发,与广告实际收益脱钩,产品确认
-# 从来不是设计内的口径。签到弹窗里的「看广告膨胀」现在与福利页看视频走同一条 reward_video
-# 路径(按 eCPM 公式发),奖励只剩「签到」+「看视频」两种。历史发币流水(coin_transaction
-# .biz_type='signin_boost')保留不动——钱是真发过的,账必须留。
-SUPPORTED_REWARD_SCENES = {REWARD_SCENE_REWARD_VIDEO}
+SUPPORTED_REWARD_SCENES = {REWARD_SCENE_REWARD_VIDEO, REWARD_SCENE_SIGNIN_BOOST}
 
 
 def _parse_extra(raw_extra: str | None) -> dict[str, str]:
@@ -123,11 +118,6 @@ def pangle_callback(request: Request, db: DbSession) -> PangleCallbackOut:
         extra.update(_parse_extra(params.get(extra_key)))
     reward_scene = extra.get("reward_scene") or REWARD_SCENE_REWARD_VIDEO
     ad_session_id = extra.get("ad_session_id")
-    # 「这条广告属于哪一轮膨胀」。纯标签:不参与发奖判定,只让 reward-result 能把同一轮求和成
-    # 弹窗要显示的累计值(见 crud_ad.round_coin_total)。老客户端不带 → NULL → 累计值返 null。
-    boost_round_id = (extra.get("boost_round_id") or None)
-    if boost_round_id is not None:
-        boost_round_id = boost_round_id[:64]
     ecpm = params.get("ecpm")
 
     # 环境隔离:激励视频 mediaExtra 里带「这次观看属于哪个后端环境」(srv_env=dev/prod,客户端按
@@ -179,11 +169,50 @@ def pangle_callback(request: Request, db: DbSession) -> PangleCallbackOut:
                 user_id, trans_id, reward_scene,
             )
             return PangleCallbackOut(is_verify=False, reason=REASON_BAD_PARAMS)
-        rec = crud_ad.grant_ad_reward(
-            db, user_id, trans_id, ecpm=ecpm, ad_session_id=ad_session_id,
-            reward_scene=REWARD_SCENE_REWARD_VIDEO, boost_round_id=boost_round_id,
-            reward_name=params.get("reward_name"), raw=raw[:1024],
-        )
+        if reward_scene == REWARD_SCENE_SIGNIN_BOOST:
+            try:
+                boost, _balance = crud_signin.boost_today_signin(
+                    db, user_id, ad_ref_id=trans_id, commit=False
+                )
+            except crud_signin.NotSignedTodayError:
+                db.rollback()
+                rec = crud_ad.record_external_reward(
+                    db, user_id, trans_id, coin=0, reward_scene=reward_scene,
+                    ad_session_id=ad_session_id, ecpm=ecpm,
+                    reward_name=params.get("reward_name"), raw=raw[:1024],
+                    status="not_signed",
+                )
+            except crud_signin.AlreadyBoostedError:
+                db.rollback()
+                rec = crud_ad.record_external_reward(
+                    db, user_id, trans_id, coin=0, reward_scene=reward_scene,
+                    ad_session_id=ad_session_id, ecpm=ecpm,
+                    reward_name=params.get("reward_name"), raw=raw[:1024],
+                    status="already_boosted",
+                )
+            except crud_signin.LastCycleDayBoostBlockedError:
+                db.rollback()
+                rec = crud_ad.record_external_reward(
+                    db, user_id, trans_id, coin=0, reward_scene=reward_scene,
+                    ad_session_id=ad_session_id, ecpm=ecpm,
+                    reward_name=params.get("reward_name"), raw=raw[:1024],
+                    status="last_day",
+                )
+            else:
+                rec = crud_ad.record_external_reward(
+                    db, user_id, trans_id, coin=boost.coin_awarded,
+                    reward_scene=reward_scene, ad_session_id=ad_session_id, ecpm=ecpm,
+                    reward_name=params.get("reward_name"), raw=raw[:1024],
+                    commit=False,
+                )
+                db.commit()
+                db.refresh(rec)
+        else:
+            rec = crud_ad.grant_ad_reward(
+                db, user_id, trans_id, ecpm=ecpm, ad_session_id=ad_session_id,
+                reward_scene=REWARD_SCENE_REWARD_VIDEO,
+                reward_name=params.get("reward_name"), raw=raw[:1024],
+            )
     except crud_ad.UnknownUserError:
         logger.warning("pangle callback unknown user_id=%d trans_id=%s", user_id, trans_id)
         return PangleCallbackOut(is_verify=False, reason=REASON_UNKNOWN_USER)
@@ -213,46 +242,6 @@ def reward_status(user: CurrentUser, db: DbSession) -> AdRewardStatusOut:
     )
 
 
-@router.get(
-    "/reward-result/{ad_session_id}",
-    response_model=AdRewardResultOut,
-    summary="按 ad_session_id 查本次广告的权威发奖结果",
-    dependencies=[Depends(rate_limit(120, 60, "ad-reward-result"))],
-)
-def reward_result(
-    user: CurrentUser,
-    db: DbSession,
-    ad_session_id: str = Path(..., min_length=8, max_length=64, description="本次广告会话 id"),
-) -> AdRewardResultOut:
-    """客户端看完激励视频后轮询本接口拿**本次实发金币 + 本轮累计**用于弹窗,不再用余额差 /
-    coin_per_ad 估算(修「弹窗数值与真实金币对不上」)。
-
-    round_coin 是「恭喜累计获得奖励」弹窗真正显示的数:本轮(= 客户端的 boost_round_id)所有
-    granted 记录之和。由服务端求和而不是客户端自己累加——客户端进程被杀/重建后本地累计会丢,
-    发奖记录不会。取不到轮 id(pending / 老客户端 / extra 丢失)时为 null,客户端退回显示单条。
-
-    S2S 回调异步:查不到记录 = 回调还没到 → 返 200 + status='pending' 让客户端继续重试,
-    **不返 404**(404 只表示路由不存在)。纯只读:发奖仍只由验签过的 S2S 回调完成,
-    这里不写库、不产生任何奖励,被刷也只是查自己的记录。
-    """
-    rec = crud_ad.find_by_session(db, user.id, ad_session_id)
-    if rec is None:
-        # 连记录都没有 → 不知道属于哪一轮,round_coin 一并为 null(不是 0,0 会被当成"本轮没赚到")
-        return AdRewardResultOut(
-            ad_session_id=ad_session_id, status="pending", coin=None, round_coin=None,
-        )
-    # 本条不是 granted 时**仍返本轮累计**(这条按 0 计):第 3 条撞每日上限那下,客户端的限额
-    # toast 要显示的是前两条已到账的总额,不是空。
-    round_coin = (
-        crud_ad.round_coin_total(db, user.id, rec.boost_round_id)
-        if rec.boost_round_id
-        else None
-    )
-    return AdRewardResultOut(
-        ad_session_id=ad_session_id, status=rec.status, coin=rec.coin, round_coin=round_coin,
-    )
-
-
 @router.post(
     "/watch-report",
     response_model=WatchReportOut,
@@ -328,27 +317,55 @@ def test_grant(user: CurrentUser, db: DbSession, payload: TestGrantIn | None = N
     if reward_scene not in SUPPORTED_REWARD_SCENES:
         raise HTTPException(status_code=status.HTTP_422_UNPROCESSABLE_ENTITY, detail="bad reward_scene")
 
-    # 每次新 trans_id,模拟一次独立的穿山甲发奖回调(幂等键各不相同 → 每次都发,直到当日上限)
+    # 每次新 trans_id,模拟一次独立的穿山甲发奖回调(幂等键各不相同 → 每次都发,直到当日上限/今日膨胀一次)
     trans_id = f"test-{user.id}-{uuid.uuid4().hex}"
-    # 正式链路的轮次 id 走 S2S 的 mediaExtra;本接口不经 S2S,只能由 body 补,否则 debug 包
-    # 的 reward-result 恒返 round_coin=null,「弹窗 40 → 60」那套累计验收在本地跑不起来。
-    boost_round_id = (payload.boost_round_id if payload is not None else None) or None
-    # 优先用客户端按 ad_session_id 上报的真实 eCPM(走与正式发奖相同的公式);
-    # 取不到或 eCPM≤0(测试应用常返 0/假值)时兜底 200,保证本地联调仍能验出非零金币。
-    ad_session_id = payload.ad_session_id if payload is not None else None
-    ecpm_val = "200"
-    if ad_session_id:
-        ecpm_rec = crud_ecpm.find_by_session(db, user_id=user.id, ad_session_id=ad_session_id)
-        if ecpm_rec is not None and rewards.parse_ecpm_fen(ecpm_rec.ecpm_raw) > 0:
-            ecpm_val = ecpm_rec.ecpm_raw
-    try:
-        rec = crud_ad.grant_ad_reward(
-            db, user.id, trans_id, ecpm=ecpm_val, ad_session_id=ad_session_id,
-            boost_round_id=boost_round_id,
-            reward_name="测试发奖", raw=f"client debug test-grant ecpm={ecpm_val}",
-        )
-    except crud_ad.UnknownUserError as e:
-        raise HTTPException(status_code=status.HTTP_404_NOT_FOUND, detail="user not found") from e
+    if reward_scene == REWARD_SCENE_SIGNIN_BOOST:
+        try:
+            boost, _balance = crud_signin.boost_today_signin(
+                db, user.id, ad_ref_id=trans_id, commit=False
+            )
+        except crud_signin.NotSignedTodayError:
+            db.rollback()
+            rec = crud_ad.record_external_reward(
+                db, user.id, trans_id, coin=0, reward_scene=reward_scene,
+                raw="client debug test-grant signin_boost", status="not_signed",
+            )
+        except crud_signin.AlreadyBoostedError:
+            db.rollback()
+            rec = crud_ad.record_external_reward(
+                db, user.id, trans_id, coin=0, reward_scene=reward_scene,
+                raw="client debug test-grant signin_boost", status="already_boosted",
+            )
+        except crud_signin.LastCycleDayBoostBlockedError:
+            db.rollback()
+            rec = crud_ad.record_external_reward(
+                db, user.id, trans_id, coin=0, reward_scene=reward_scene,
+                raw="client debug test-grant signin_boost", status="last_day",
+            )
+        else:
+            rec = crud_ad.record_external_reward(
+                db, user.id, trans_id, coin=boost.coin_awarded,
+                reward_scene=reward_scene, reward_name="测试签到膨胀",
+                raw="client debug test-grant signin_boost", commit=False,
+            )
+            db.commit()
+            db.refresh(rec)
+    else:
+        # 优先用客户端按 ad_session_id 上报的真实 eCPM(走与正式发奖相同的公式);
+        # 取不到或 eCPM≤0(测试应用常返 0/假值)时兜底 200,保证本地联调仍能验出非零金币。
+        ad_session_id = payload.ad_session_id if payload is not None else None
+        ecpm_val = "200"
+        if ad_session_id:
+            ecpm_rec = crud_ecpm.find_by_session(db, user_id=user.id, ad_session_id=ad_session_id)
+            if ecpm_rec is not None and rewards.parse_ecpm_fen(ecpm_rec.ecpm_raw) > 0:
+                ecpm_val = ecpm_rec.ecpm_raw
+        try:
+            rec = crud_ad.grant_ad_reward(
+                db, user.id, trans_id, ecpm=ecpm_val, ad_session_id=ad_session_id,
+                reward_name="测试发奖", raw=f"client debug test-grant ecpm={ecpm_val}",
+            )
+        except crud_ad.UnknownUserError as e:
+            raise HTTPException(status_code=status.HTTP_404_NOT_FOUND, detail="user not found") from e
 
     (used, limit, coin_per, round_count, cooldown_until,
      _watched, _watch_limit) = crud_ad.today_status(db, user.id)

app/api/v1/auth.py

@@ -12,16 +12,11 @@ from __future__ import annotations
 
 import logging
 
-from fastapi import APIRouter, HTTPException, Request
+from fastapi import APIRouter, HTTPException, Request, status
 
 from app.api.deps import CurrentUser, DbSession
 from app.core import test_account
-from app.core.ratelimit import (
-    RateLimitRule,
-    check_rate_limits,
-    enforce_rate_limit,
-    record_rate_limits,
-)
+from app.core.ratelimit import enforce_rate_limit
 from app.core.security import TokenError, decode_token, issue_token_pair
 from app.integrations.jiguang import JiguangError, mask_phone, verify_and_get_phone
 from app.integrations.sms import SmsError, send_code, verify_code
@@ -45,10 +40,9 @@ router = APIRouter(prefix="/api/v1/auth", tags=["auth"])
 
 # 手机号登录防刷:同一设备(device_id) + 同一 IP 每小时最多的登录尝试次数(成功/失败都计)。
 SMS_LOGIN_MAX_PER_HOUR = 5
-# 发码防刷(同一设备 device_id + 同一 IP,**只按成功发码计数**;被单号 60s 冷却挡下的重发不占额度):
+# 发码防刷:同一设备(device_id) + 同一 IP 每小时最多的发码次数。
 # 堵「换手机号绕开单号 60s 冷却」的洞 —— 冷却是单号维度,一机换号能绕开。
-SMS_SEND_MAX_PER_HOUR_PER_DEVICE = 5     # 每小时上限
-SMS_SEND_MAX_PER_DAY_PER_DEVICE = 20     # 每天上限(再叠一层日封顶,挡低频长时间轰炸)
+SMS_SEND_MAX_PER_HOUR_PER_DEVICE = 5
 
 
 def _login_response(
@@ -105,26 +99,23 @@ def sms_send(req: SmsSendRequest, request: Request) -> SmsSendResponse:
         logger.info("test_account sms_send short-circuit (不真发)")
         return SmsSendResponse(sent=True, mock=True, cooldown_sec=0)
 
-    # 发码防刷:同一设备(device_id) + 同一 IP,每小时 / 每天两道闸,**均只按成功发码计数**。
-    # 补「换手机号绕开单号 60s 冷却」的洞(冷却是单号维度,一机换号能绕);设备维度按机器封顶,挡短信轰炸/烧钱。
-    # 关键:被单号 60s 冷却挡下的重发是「没真发、没烧钱」→ 不该占额度。故 check(先判)放在真发之前
-    # (超限直接 429、不真发),record(计数)只在 send_code 成功后调 —— 冷却/供应商失败抛 429 时直接返回、不计数。
-    send_rules = [
-        RateLimitRule("sms-send-device", SMS_SEND_MAX_PER_HOUR_PER_DEVICE, 3600,
-                      "操作过于频繁,请稍后再试"),
-        RateLimitRule("sms-send-device-daily", SMS_SEND_MAX_PER_DAY_PER_DEVICE, 86400,
-                      "今日验证码发送次数过多,请明天再试"),
-    ]
-    check_rate_limits(request, subject=req.device_id, rules=send_rules)
+    # 防刷:同一设备(device_id) + 同一 IP 每小时最多 SMS_SEND_MAX_PER_HOUR_PER_DEVICE 次发码。
+    # 补「换手机号绕开单号 60s 冷却」的洞(冷却是单号维度,一机换号能绕);设备维度按机器封顶,
+    # 挡短信轰炸/烧钱。放在真发(send_code)之前 → 超限直接拦下、不真发短信。
+    enforce_rate_limit(
+        request,
+        scope="sms-send-device",
+        subject=req.device_id,
+        limit=SMS_SEND_MAX_PER_HOUR_PER_DEVICE,
+        window_sec=3600,
+        detail="操作过于频繁,请稍后再试",
+    )
 
     try:
         cooldown = send_code(req.phone)
     except SmsError as e:
         raise HTTPException(status_code=e.status_code, detail=str(e)) from e
 
-    # 发码成功 → 两道闸各 +1(被单号冷却挡下的重发走不到这里,故不占额度)
-    record_rate_limits(request, subject=req.device_id, rules=send_rules)
-
     from app.core.config import settings  # 局部 import 避免循环
 
     return SmsSendResponse(sent=True, mock=settings.SMS_MOCK, cooldown_sec=cooldown)

app/api/v1/compare_record.py

@@ -27,7 +27,6 @@ from app.schemas.compare_record import (
     ComparisonRecordOut,
     ComparisonRecordPage,
 )
-from app.services.llm_cost import compute_llm_cost, get_llm_prices
 from app.services.pricebot_llm_calls import fetch_llm_calls
 
 logger = logging.getLogger("shagua.compare_record")
@@ -82,8 +81,6 @@ def _backfill_llm_calls(record_id: int, trace_id: str) -> None:
         # error 的调用 usage 可能为 None,or {} 兜底)
         rec.input_tokens = sum((c.get("usage") or {}).get("prompt_tokens") or 0 for c in calls)
         rec.output_tokens = sum((c.get("usage") or {}).get("completion_tokens") or 0 for c in calls)
-        # 本次比价 LLM 成本(元)+ 当时单价快照:按 app_config 现价逐模型算好冻结(services/llm_cost.py)。
-        rec.llm_cost_yuan, rec.llm_price_snapshot = compute_llm_cost(calls, get_llm_prices(db))
         db.commit()
         logger.info(
             "backfill llm_calls trace=%s n=%d in_tok=%d out_tok=%d",

app/api/v1/signin.py

@@ -1,12 +1,9 @@
 """签到 endpoint。
 
 路由前缀 `/api/v1/signin`:
-  GET  /status    今日签到状态 + 7 天档位
+  GET  /status    今日签到状态 + 14 天档位
   POST /          执行今日签到
-
-2026-07 下线 `POST /boost`(签到膨胀):它按固定 3000 金币补发、与广告实际收益脱钩。
-签到弹窗里的「看广告膨胀」改与福利页看视频走同一条 reward_video 路径(按 eCPM 发,
-`/ad/pangle-callback` → `/ad/reward-result` 取金额),奖励只剩「签到」+「看视频」两种。
+  POST /boost     签到后看广告膨胀金币
 """
 from __future__ import annotations
 
@@ -15,8 +12,15 @@ import logging
 from fastapi import APIRouter, HTTPException, status
 
 from app.api.deps import CurrentUser, DbSession
+from app.repositories import ad_reward as crud_ad
 from app.repositories import signin as crud_signin
-from app.schemas.welfare import SigninResultOut, SigninStatusOut
+from app.repositories import wallet as crud_wallet
+from app.schemas.welfare import (
+    SigninBoostRequest,
+    SigninBoostResultOut,
+    SigninResultOut,
+    SigninStatusOut,
+)
 
 logger = logging.getLogger("shagua.signin")
 
@@ -46,3 +50,34 @@ def do_signin(user: CurrentUser, db: DbSession) -> SigninResultOut:
         streak=record.streak,
         coin_balance=balance,
     )
+
+
+@router.post("/boost", response_model=SigninBoostResultOut, summary="签到后看广告膨胀金币")
+def boost_signin(
+    payload: SigninBoostRequest, user: CurrentUser, db: DbSession
+) -> SigninBoostResultOut:
+    if not payload.ad_ref_id:
+        raise HTTPException(status_code=status.HTTP_409_CONFLICT, detail="ad reward required")
+    ad_rec = crud_ad.find_by_trans(db, payload.ad_ref_id)
+    if (
+        ad_rec is None
+        or ad_rec.user_id != user.id
+        or ad_rec.reward_scene != "signin_boost"
+        or ad_rec.status != "granted"
+    ):
+        raise HTTPException(status_code=status.HTTP_409_CONFLICT, detail="ad reward not verified")
+    record = crud_signin.boost_by_ad_ref(db, user.id, payload.ad_ref_id)
+    if record is None:
+        raise HTTPException(status_code=status.HTTP_409_CONFLICT, detail="signin boost not granted")
+    acc = crud_wallet.get_or_create_account(db, user.id)
+    balance = acc.coin_balance
+
+    logger.info(
+        "signin boost ok user_id=%d date=%s coin=%d",
+        user.id, record.signin_date, record.coin_awarded,
+    )
+    return SigninBoostResultOut(
+        coin_awarded=record.coin_awarded,
+        coin_balance=balance,
+        signin_date=record.signin_date.isoformat(),
+    )

app/api/v1/wallet.py

@@ -201,7 +201,7 @@ def withdraw_info(
     "/withdraw",
     response_model=WithdrawResultOut,
     summary="发起提现(扣款建单,待人工审核;审核通过后才打款)",
-    dependencies=[Depends(rate_limit(5, 60, "withdraw"))],  # IP 级粗限流;用户级未完成单限制在仓库层
+    dependencies=[Depends(rate_limit(5, 60, "withdraw"))],  # IP 级粗限流;余额/档位/幂等限制在仓库层
 )
 def withdraw(req: WithdrawRequest, user: CurrentUser, db: DbSession) -> WithdrawResultOut:
     # 提现发起本身不调微信(打款在审核通过后),但仍要求微信支付已配置——否则审核通过也打不了款,提前拦
@@ -222,11 +222,6 @@ def withdraw(req: WithdrawRequest, user: CurrentUser, db: DbSession) -> Withdraw
         ) from e
     except crud_wallet.WechatNotBoundError as e:
         raise HTTPException(status_code=status.HTTP_400_BAD_REQUEST, detail="请先绑定微信") from e
-    except crud_wallet.WithdrawTooFrequentError as e:
-        raise HTTPException(
-            status_code=status.HTTP_409_CONFLICT,
-            detail="已有提现申请正在审核或打款中,请处理完成后再申请",
-        ) from e
     except crud_wallet.WithdrawTierUnavailableError as e:
         # 福利页档位闸(7-9):次数满/已选其他额度。正常客户端已按 tiers 预拦,此处兜底防绕过。
         raise HTTPException(status_code=status.HTTP_409_CONFLICT, detail="今日额度已达上限") from e

app/core/config_schema.py

@@ -66,6 +66,11 @@ CONFIG_DEFS: dict[str, dict[str, Any]] = {
         "default": r.VIDEO_ROUND_COOLDOWN_SECONDS, "label": "广告关闭后冷却(秒)",
         "group": "看广告", "type": "int", "help": "点击退出广告后,下次点击观看前的冷却时间,默认 3 秒。",
     },
+    "signin_boost_coin": {
+        "default": r.SIGNIN_BOOST_COIN, "label": "签到膨胀固定金币",
+        "group": "签到", "type": "int",
+        "help": "Day1-Day6 签到后看完激励视频额外发放的固定金币;Day7 不展示也不允许膨胀。",
+    },
     "comparing_ad_enabled": {
         "default": True, "label": "比价/领券期信息流广告",
         "group": "看广告", "type": "bool", "hidden": True,
@@ -91,19 +96,4 @@ CONFIG_DEFS: dict[str, dict[str, Any]] = {
         "group": "首页轮播", "type": "enum", "hidden": True,
         "help": "mixed=真实优先+种子补位(默认);real=只用真实比价记录;seed=只用种子/合成(演示)。",
     },
-    # 比价 LLM 调用成本计价。值是嵌套 JSON(非 str→int),借 dict_str_int 类型在配置页走原始 JSON
-    # 编辑框;set_value 不校验类型,嵌套 JSON 照存。
-    "llm_token_price": {
-        "default": {
-            "per_model": {"qwen3.5-flash": {"input_per_1m": 0.8, "output_per_1m": 2.0}},
-            "default": {"input_per_1m": 3.0, "output_per_1m": 15.0},
-            "currency": "CNY", "unit": "per_1m_tokens",
-        },
-        "label": "LLM 模型单价(元/百万 token)",
-        "group": "LLM 成本", "type": "dict_str_int",
-        "help": (
-            "比价 LLM 调用成本计价。JSON:per_model 按模型配 input/output 单价(元/1M token),"
-            "default 兜底未登记的模型。改价只影响之后回填的新记录,历史记录用当时价格快照。"
-        ),
-    },
 }

app/core/ratelimit.py

@@ -9,41 +9,29 @@ from __future__ import annotations
 
 import threading
 import time
-from typing import NamedTuple
 
 from fastapi import HTTPException, Request, status
 
 from app.core.config import settings
 
-# key -> (window_start_ts, count, window_sec)
-# 存每个 key 自己的 window_sec:_buckets 混着不同窗口(60s 广告 / 3600s 登录 / 86400s 日闸)的 key,
-# GC 必须按各 key 自己的窗口判过期(见 [_purge_expired]),否则短窗口调用触发的 GC 会误删长窗口 key。
-_buckets: dict[str, tuple[float, int, float]] = {}
+# key -> (window_start_ts, count)
+_buckets: dict[str, tuple[float, int]] = {}
 _lock = threading.Lock()
-_GC_THRESHOLD = 10000  # _buckets 超此阈值才顺手清过期 key(仿 sms.py;测试可 monkeypatch 调小强制每次扫)
-
-
-def _purge_expired(now: float) -> None:
-    """清过期 key(**仅在持有 _lock 时调用**)。按每个 key 自己存的 window_sec 判过期,而非调用方的窗口
-    —— _buckets 是全局共享、混着 60s(广告)/3600s(登录)/86400s(日闸)不同窗口的 key;若用调用方窗口,
-    高频的 60s 广告端点触发 GC 时会把本该活 3600s/86400s 的登录/日闸计数一并删掉,使其在规模上(超阈值才
-    触发本清理)被反复清零而失效。仅在超阈值时扫,低频、开销可忽略。"""
-    if len(_buckets) <= _GC_THRESHOLD:
-        return
-    for k in [k for k, (s, _, w) in _buckets.items() if now - s >= w]:
-        _buckets.pop(k, None)
 
 
 def _hit(key: str, limit: int, window_sec: float) -> bool:
     """记一次访问。返回 True=放行,False=超限。"""
     now = time.monotonic()
     with _lock:
-        start, count, _ = _buckets.get(key, (now, 0, window_sec))
+        start, count = _buckets.get(key, (now, 0))
         if now - start >= window_sec:  # 窗口过期,重置
             start, count = now, 0
         count += 1
-        _buckets[key] = (start, count, window_sec)
-        _purge_expired(now)  # 顺手清过期 key(按各自窗口),防内存无限涨
+        _buckets[key] = (start, count)
+        # 顺手清理过期 key,防内存无限涨(低频访问足够)
+        if len(_buckets) > 10000:
+            for k in [k for k, (s, _) in _buckets.items() if now - s >= window_sec]:
+                _buckets.pop(k, None)
     return count <= limit
 
 
@@ -95,76 +83,3 @@ def enforce_rate_limit(
             status_code=status.HTTP_429_TOO_MANY_REQUESTS,
             detail=detail,
         )
-
-
-# ===================== 先判 / 后记(只按「成功」计数)=====================
-# _hit 是原子「判+记」:一调用就 +1,适合登录爆破(失败尝试也要计)。但对「短信发码」这类
-# **只想给成功动作计数**的场景不合适 —— 被单号冷却挡下的重发没真发、没烧钱,不该占额度。
-# 故拆成 _peek(只判不记)+ _commit(只记):check_rate_limits 先判 → 动作 → 成功后 record。
-
-
-class RateLimitRule(NamedTuple):
-    """一条限流规则。scope 区分不同闸(不同 key 前缀);同一 (subject, IP) 在 window_sec
-    内最多 limit 次,超限抛 429 用 detail 文案。
-
-    (scope, window_sec) 成对绑在一条规则里 —— check(先判)与 record(计数)复用同一条,
-    避免两处把窗口/scope 写歪导致 key 对不上。
-    """
-
-    scope: str
-    limit: int
-    window_sec: float
-    detail: str = "操作过于频繁,请稍后再试"
-
-
-def _peek(key: str, limit: int, window_sec: float) -> bool:
-    """只读:当前窗口内是否还没到上限(count < limit)。**不改计数**。
-    与 [_commit] 配对实现「先判后记」——只在动作成功后才 _commit。"""
-    now = time.monotonic()
-    with _lock:
-        start, count, _ = _buckets.get(key, (now, 0, window_sec))
-        if now - start >= window_sec:  # 窗口已过期 → 视作已重置(count 归零)
-            count = 0
-    return count < limit
-
-
-def _commit(key: str, window_sec: float) -> None:
-    """记一次访问(+1)。窗口过期则以本次为起点重置。仅在动作成功后调用。"""
-    now = time.monotonic()
-    with _lock:
-        start, count, _ = _buckets.get(key, (now, 0, window_sec))
-        if now - start >= window_sec:  # 窗口过期,重置
-            start, count = now, 0
-        _buckets[key] = (start, count + 1, window_sec)
-        _purge_expired(now)  # 顺手清过期 key(按各自窗口,同 [_hit])
-
-
-def check_rate_limits(request: Request, subject: str, rules: list[RateLimitRule]) -> None:
-    """【先判】一组限流:任一规则已达上限即抛 429,且**不改计数**。
-
-    配合 [record_rate_limits] 实现「只按成功计数」:先 check 所有闸(全未超才继续)→ 执行动作
-    → 动作**成功后**再 record。动作被下游挡下(如短信单号冷却)、没真正发生时不 record → 不占额度。
-    key = `scope:subject:client_ip`(与 [enforce_rate_limit] 同款)。
-    """
-    if not settings.RATE_LIMIT_ENABLED:
-        return
-    ip = _client_ip(request)
-    for rule in rules:
-        if not _peek(f"{rule.scope}:{subject}:{ip}", rule.limit, rule.window_sec):
-            raise HTTPException(
-                status_code=status.HTTP_429_TOO_MANY_REQUESTS,
-                detail=rule.detail,
-            )
-
-
-def record_rate_limits(request: Request, subject: str, rules: list[RateLimitRule]) -> None:
-    """【记一次】一组限流(每条规则 +1)。仅在动作成功后调用,与 [check_rate_limits] 配对。
-
-    ⚠️ check→动作→record 非原子:并发突发下计数可能略超 limit(每个在途请求各 +1)。对
-    「防脚本/防轰炸」的安全网定位可接受;要精确配额需迁 Redis(见模块 docstring)。
-    """
-    if not settings.RATE_LIMIT_ENABLED:
-        return
-    ip = _client_ip(request)
-    for rule in rules:
-        _commit(f"{rule.scope}:{subject}:{ip}", rule.window_sec)

app/core/rewards.py

@@ -239,8 +239,8 @@ def calculate_ad_reward_coin(ecpm: str | int | float | None, count_after_this: i
     return max(0, round(yuan * COIN_PER_YUAN))
 
 
-# 签到膨胀(SIGNIN_BOOST_COIN,固定 3000)已于 2026-07 下线:它与广告实际收益脱钩,产品确认
-# 非设计内口径。签到弹窗的「看广告膨胀」现与福利页看视频同走 calculate_ad_reward_coin。
+# 签到看广告膨胀:S2S 固定补发(原型 2026-06 由 2000 提到 3000,对应 CTA「看广告最高膨胀至3000金币」)。
+SIGNIN_BOOST_COIN: int = 3000
 
 
 # ===== 看激励视频发金币(穿山甲 S2S 服务端回调发奖)=====
@@ -334,3 +334,5 @@ def get_ad_cooldown_sec(db) -> int:  # noqa: ANN001
     return int(_cfg(db, "ad_cooldown_sec"))
 
 
+def get_signin_boost_coin(db) -> int:  # noqa: ANN001
+    return int(_cfg(db, "signin_boost_coin"))

app/integrations/sms.py

@@ -13,8 +13,7 @@ worker / 多机时内存不共享 → 冷却、校验都会失效,届时迁移
 
 防刷两层(短信花钱 + `/sms/send` 在登录前无法 JWT 鉴权):
   1. 单号 `SMS_SEND_INTERVAL_SEC` 冷却(本文件)
-  2. 单设备(device_id)+ IP 每小时 / 每天频控(api 层 auth.sms_send 的 check/record_rate_limits,
-     **只按成功发码计数** —— 被本文件单号冷却挡下的重发不占额度)+ 极光控制台 IP 白名单/防轰炸(运维侧)。
+  2. 单设备(device_id)每小时频控(api 层 auth.sms_send 内 enforce_rate_limit)+ 极光控制台 IP 白名单/防轰炸(运维侧)。
      ⚠️ 原「单 IP 频控(rate_limit 依赖)」2026-06-26 按产品要求删除、改设备维度;但 device_id 客户端可伪造/轮换,
         脚本轮换 id 能绕过本层 → 挡脚本狂发主要靠极光控制台侧(+ 可选 nginx 限流)。
      ⚠️ 原「单号每日上限」2026-07-03 按精简要求删除(mentor 定:登录风控只留单号冷却 + 单设备频控);

app/models/__init__.py

@@ -37,7 +37,7 @@ from app.models.ops_stat_config import OpsStatConfig  # noqa: F401
 from app.models.price_observation import PriceObservation  # noqa: F401
 from app.models.price_report import PriceReport  # noqa: F401
 from app.models.savings import SavingsRecord  # noqa: F401
-from app.models.signin import SigninRecord  # noqa: F401
+from app.models.signin import SigninBoostRecord, SigninRecord  # noqa: F401
 from app.models.store_mapping import StoreMapping  # noqa: F401
 from app.models.task import UserTask  # noqa: F401
 from app.models.user import User  # noqa: F401

app/models/ad_reward.py

@@ -8,7 +8,7 @@ from __future__ import annotations
 
 from datetime import datetime
 
-from sqlalchemy import DateTime, ForeignKey, Index, Integer, String, func
+from sqlalchemy import DateTime, ForeignKey, Integer, String, func
 from sqlalchemy.orm import Mapped, mapped_column
 
 from app.db.base import Base
@@ -16,10 +16,6 @@ from app.db.base import Base
 
 class AdRewardRecord(Base):
     __tablename__ = "ad_reward_record"
-    __table_args__ = (
-        # 「本轮膨胀累计发了多少」= SUM(coin) WHERE user_id=? AND boost_round_id=? AND status='granted'
-        Index("ix_ad_reward_user_boost_round", "user_id", "boost_round_id"),
-    )
 
     id: Mapped[int] = mapped_column(Integer, primary_key=True, autoincrement=True)
     # 穿山甲交易号,幂等键(同号回调不重复发奖)
@@ -35,10 +31,6 @@ class AdRewardRecord(Base):
     reward_scene: Mapped[str] = mapped_column(String(32), nullable=False, default="reward_video")
     # 客户端生成并通过 extra 透传的广告会话 id
     ad_session_id: Mapped[str | None] = mapped_column(String(64), index=True, nullable=True)
-    # 客户端生成并通过 extra 透传的「膨胀轮」id:一轮 = 用户点「去膨胀」到点「放弃赚钱」之间连看的
-    # 若干条广告。纯标签,不影响发多少/发不发,只用于把同一轮的发奖记录求和成弹窗要显示的累计值。
-    # 轮次边界完全由客户端定(它才知道用户点了放弃);老客户端/extra 丢失时为 NULL → 累计值返 null。
-    boost_round_id: Mapped[str | None] = mapped_column(String(64), nullable=True)
     # 本次发奖采用的 eCPM 原始值(回调自带或按 ad_session_id 匹配的客户端上报)
     ecpm_raw: Mapped[str | None] = mapped_column(String(32), nullable=True)
     # 来源(广告收益报表用):我们的应用环境 prod/test + 我们配置的代码位 104xxx。

app/models/comparison.py

@@ -137,12 +137,6 @@ class ComparisonRecord(Base):
     # 每次 LLM 调用明细 [{scene,model,input_messages,output,usage,latency_ms,error}];
     # server 收上报后按 trace_id 同机拉 pricebot 落库(见 compare_record 端点)。旧记录/未采集为 None。
     llm_calls: Mapped[list | None] = mapped_column(_JSON, nullable=True)
-    # 本次比价 LLM 总成本(元):回填时按「当时的价」逐模型算好冻结(见 services/llm_cost.py)。
-    # 单次亚分级 → float「元」(不用 *_cents)。旧记录/未回填为 None,前端回退「估算成本」。
-    llm_cost_yuan: Mapped[float | None] = mapped_column(Float, nullable=True)
-    # 算成本所用单价快照 {mode, prices:{model:{input_per_1m,output_per_1m,_source}}}:app_config 只存
-    # 当前价、不留历史,故把当时价冻结进来供审计/复算。
-    llm_price_snapshot: Mapped[dict | None] = mapped_column(_JSON, nullable=True)
 
     created_at: Mapped[datetime] = mapped_column(
         DateTime(timezone=True), server_default=func.now(), index=True, nullable=False

app/models/signin.py

@@ -1,10 +1,6 @@
 """签到记录表。
 
 每次签到一行,(user_id, signin_date) 唯一,天然防一天签两次。
-
-2026-07 下线 `signin_boost_record`(签到膨胀):膨胀按固定 3000 金币补发、与广告实际收益
-脱钩,产品确认非设计内口径。签到弹窗的「看广告膨胀」改走 reward_video(按 eCPM 发,记在
-`ad_reward_record`)。历史发币流水 `coin_transaction.biz_type='signin_boost'` 保留不动。
   - cycle_day: 1..7,7 天循环里今天落在第几档,决定发多少金币;断签后重置回 1
               (周期长度 = rewards.SIGNIN_CYCLE_LEN,2026-06 由 14 天改 7 天一轮)。
   - streak:   连续签到天数(不封顶),用于"已连续签到 N 天"展示;断签后重置回 1。
@@ -13,7 +9,7 @@ from __future__ import annotations
 
 from datetime import date, datetime
 
-from sqlalchemy import Date, DateTime, ForeignKey, Integer, UniqueConstraint, func
+from sqlalchemy import Date, DateTime, ForeignKey, Integer, String, UniqueConstraint, func
 from sqlalchemy.orm import Mapped, mapped_column
 
 from app.db.base import Base
@@ -46,3 +42,29 @@ class SigninRecord(Base):
         )
 
 
+class SigninBoostRecord(Base):
+    """签到后看广告膨胀记录。
+
+    一天最多膨胀一次,补发金额等于当天签到原始奖励。独立表用于防并发重复补发,
+    后续接入真实 S2S 广告 session 时可把 ad_ref_id 回填为广告会话/交易号。
+    """
+
+    __tablename__ = "signin_boost_record"
+    __table_args__ = (
+        UniqueConstraint("user_id", "signin_date", name="uq_signin_boost_user_date"),
+    )
+
+    id: Mapped[int] = mapped_column(Integer, primary_key=True, autoincrement=True)
+    user_id: Mapped[int] = mapped_column(
+        Integer, ForeignKey("user.id"), index=True, nullable=False
+    )
+    signin_date: Mapped[date] = mapped_column(Date, nullable=False)
+    coin_awarded: Mapped[int] = mapped_column(Integer, nullable=False)
+    ad_ref_id: Mapped[str | None] = mapped_column(String(64), nullable=True)
+
+    created_at: Mapped[datetime] = mapped_column(
+        DateTime(timezone=True), server_default=func.now(), nullable=False
+    )
+
+    def __repr__(self) -> str:  # pragma: no cover
+        return f"<SigninBoostRecord user_id={self.user_id} date={self.signin_date} coin={self.coin_awarded}>"

app/models/wallet.py

@@ -96,16 +96,6 @@ class WithdrawOrder(Base):
     """
 
     __tablename__ = "withdraw_order"
-    __table_args__ = (
-        Index(
-            "ux_withdraw_order_user_active",
-            "user_id",
-            unique=True,
-            sqlite_where=text("status IN ('reviewing', 'pending')"),
-            postgresql_where=text("status IN ('reviewing', 'pending')"),
-        ),
-    )
-
     id: Mapped[int] = mapped_column(Integer, primary_key=True, autoincrement=True)
     user_id: Mapped[int] = mapped_column(
         Integer, ForeignKey("user.id"), index=True, nullable=False

app/repositories/ad_reward.py

@@ -41,61 +41,6 @@ def find_by_trans(db: Session, trans_id: str) -> AdRewardRecord | None:
     return _find_by_trans(db, trans_id)
 
 
-def find_by_session(db: Session, user_id: int, ad_session_id: str) -> AdRewardRecord | None:
-    """按广告会话 id 查该用户本次广告的发奖记录,供客户端轮询弹窗金额(reward-result)。
-
-    同一 ad_session_id 可能命中多条,**必须显式优先 granted**,不能只取最近一条:
-      - 客户端先上报 closed_early、S2S 随后才姗姗来迟 → 两条,granted 反而是后写的;
-      - record_reward_noshow 只在写入前查 granted,挡不住这种后到的竞态;
-      - 本地联调重复调 test-grant → 同 session 多条 granted(trans_id 各不相同)。
-    granted 是唯一「真发了钱」的状态,取它才是权威金额;都没有再取最近一条,让客户端
-    知道没发的原因(capped/closed_early…)。按 user_id 收窄,防止拿别人的 session 探测结果。
-    """
-    granted = db.execute(
-        select(AdRewardRecord)
-        .where(
-            AdRewardRecord.user_id == user_id,
-            AdRewardRecord.ad_session_id == ad_session_id,
-            AdRewardRecord.status == "granted",
-        )
-        .order_by(AdRewardRecord.created_at.desc())
-        .limit(1)
-    ).scalars().first()
-    if granted is not None:
-        return granted
-    return db.execute(
-        select(AdRewardRecord)
-        .where(
-            AdRewardRecord.user_id == user_id,
-            AdRewardRecord.ad_session_id == ad_session_id,
-        )
-        .order_by(AdRewardRecord.created_at.desc())
-        .limit(1)
-    ).scalars().first()
-
-
-def round_coin_total(db: Session, user_id: int, boost_round_id: str) -> int:
-    """本轮膨胀累计已发金币 = 该轮所有 granted 记录的 coin 之和(含刚发的这条)。
-
-    客户端弹窗要显示的就是它:第 1 条弹 40、第 2 条弹 60(=40+20),点「放弃赚钱」后余额涨 60,
-    三个数必须相等。之所以由服务端求和而不是客户端自己累加——客户端进程被杀/低内存重建后
-    本地累计就丢了,而发奖记录不会丢。
-
-    **必须带 user_id**:boost_round_id 是客户端生成的,不带 user_id 就等于让任何人拿别人的
-    轮 id 查别人发了多少。未发奖的状态(capped/closed_early/ecpm_missing)coin 本就是 0,
-    这里按 status 过滤只是让意图显式。
-    """
-    return int(
-        db.execute(
-            select(func.coalesce(func.sum(AdRewardRecord.coin), 0)).where(
-                AdRewardRecord.user_id == user_id,
-                AdRewardRecord.boost_round_id == boost_round_id,
-                AdRewardRecord.status == "granted",
-            )
-        ).scalar_one()
-    )
-
-
 def _granted_today(db: Session, user_id: int, reward_date: str) -> int:
     return db.execute(
         select(func.count())
@@ -133,12 +78,8 @@ def grant_ad_reward(
     reward_scene: str = "reward_video",
     reward_name: str | None = None,
     raw: str | None = None,
-    boost_round_id: str | None = None,
 ) -> AdRewardRecord:
-    """福利页激励视频发奖(幂等 + 每日限额 + eCPM 公式)。
-
-    boost_round_id 只是随记录存下的标签(见 round_coin_total),**不参与任何发奖判定**。
-    """
+    """福利页激励视频发奖(幂等 + 每日限额 + eCPM 公式)。"""
     # #2 幂等:同 trans_id 已处理过 → 原样返回,不重复发
     existing = _find_by_trans(db, trans_id)
     if existing is not None:
@@ -171,7 +112,7 @@ def grant_ad_reward(
             trans_id=trans_id, user_id=user_id, coin=0, status="capped",
             reward_date=today, reward_name=reward_name, raw=raw,
             reward_scene=reward_scene, ad_session_id=ad_session_id, ecpm_raw=ecpm,
-            app_env=src_app_env, our_code_id=src_code_id, boost_round_id=boost_round_id,
+            app_env=src_app_env, our_code_id=src_code_id,
         )
         return _commit_record(db, rec, trans_id)
 
@@ -182,7 +123,7 @@ def grant_ad_reward(
             trans_id=trans_id, user_id=user_id, coin=0, status="ecpm_missing",
             reward_date=today, reward_name=reward_name, raw=raw,
             reward_scene=reward_scene, ad_session_id=ad_session_id, ecpm_raw=None,
-            app_env=src_app_env, our_code_id=src_code_id, boost_round_id=boost_round_id,
+            app_env=src_app_env, our_code_id=src_code_id,
         )
         return _commit_record(db, rec, trans_id)
 
@@ -199,7 +140,7 @@ def grant_ad_reward(
         trans_id=trans_id, user_id=user_id, coin=coin, status="granted",
         reward_date=today, reward_name=reward_name, raw=raw,
         reward_scene=reward_scene, ad_session_id=ad_session_id, ecpm_raw=ecpm_raw,
-        app_env=src_app_env, our_code_id=src_code_id, boost_round_id=boost_round_id,
+        app_env=src_app_env, our_code_id=src_code_id,
     )
     return _commit_record(db, rec, trans_id)
 
@@ -265,13 +206,8 @@ def record_external_reward(
     raw: str | None = None,
     status: str = "granted",
     commit: bool = True,
-    boost_round_id: str | None = None,
 ) -> AdRewardRecord:
-    """记录非普通看视频场景的 S2S 回调幂等,发币由调用方业务仓储完成。
-
-    boost_round_id 同 grant_ad_reward:纯标签。签到膨胀场景的 coin 也会计入本轮累计
-    (它的 coin 就是实发额),所以这里也要存,否则一轮里混了膨胀就会漏算。
-    """
+    """记录非普通看视频场景的 S2S 回调幂等,发币由调用方业务仓储完成。"""
     existing = _find_by_trans(db, trans_id)
     if existing is not None:
         return existing
@@ -288,7 +224,6 @@ def record_external_reward(
         reward_scene=reward_scene,
         ad_session_id=ad_session_id,
         ecpm_raw=ecpm,
-        boost_round_id=boost_round_id,
     )
     db.add(rec)
     if commit:

app/repositories/signin.py

@@ -11,11 +11,12 @@ from dataclasses import dataclass
 from datetime import timedelta
 
 from sqlalchemy import select
+from sqlalchemy.exc import IntegrityError
 from sqlalchemy.orm import Session
 
 from app.core import rewards
 from app.core.rewards import SIGNIN_CYCLE_LEN, cn_today
-from app.models.signin import SigninRecord
+from app.models.signin import SigninBoostRecord, SigninRecord
 from app.repositories import wallet as crud_wallet
 
 
@@ -23,6 +24,18 @@ class AlreadySignedError(Exception):
     """今天已经签过了。"""
 
 
+class NotSignedTodayError(Exception):
+    """今天尚未签到,不能膨胀。"""
+
+
+class AlreadyBoostedError(Exception):
+    """今天签到奖励已经膨胀过。"""
+
+
+class LastCycleDayBoostBlockedError(Exception):
+    """循环最后一天(第 SIGNIN_CYCLE_LEN 天)不允许签到膨胀。"""
+
+
 @dataclass
 class SigninStep:
     day: int  # 1..14
@@ -128,3 +141,69 @@ def do_signin(db: Session, user_id: int) -> tuple[SigninRecord, int]:
     db.commit()
     db.refresh(record)
     return record, acc.coin_balance
+
+
+def _today_record(db: Session, user_id: int) -> SigninRecord | None:
+    today = cn_today()
+    return db.execute(
+        select(SigninRecord).where(
+            SigninRecord.user_id == user_id,
+            SigninRecord.signin_date == today,
+        )
+    ).scalar_one_or_none()
+
+
+def boost_by_ad_ref(
+    db: Session, user_id: int, ad_ref_id: str
+) -> SigninBoostRecord | None:
+    """按广告交易号查签到膨胀记录。S2S 发奖后客户端确认用。"""
+    return db.execute(
+        select(SigninBoostRecord).where(
+            SigninBoostRecord.user_id == user_id,
+            SigninBoostRecord.ad_ref_id == ad_ref_id,
+        )
+    ).scalar_one_or_none()
+
+
+def boost_today_signin(
+    db: Session, user_id: int, *, ad_ref_id: str | None = None, commit: bool = True
+) -> tuple[SigninBoostRecord, int]:
+    """签到后看广告膨胀:固定补发配置金币。返回 (膨胀记录, 补发后余额)。"""
+    record = _today_record(db, user_id)
+    if record is None:
+        raise NotSignedTodayError
+    if record.cycle_day == SIGNIN_CYCLE_LEN:
+        raise LastCycleDayBoostBlockedError
+
+    today = record.signin_date
+    existing = db.execute(
+        select(SigninBoostRecord).where(
+            SigninBoostRecord.user_id == user_id,
+            SigninBoostRecord.signin_date == today,
+        )
+    ).scalar_one_or_none()
+    if existing is not None:
+        raise AlreadyBoostedError
+
+    boost = SigninBoostRecord(
+        user_id=user_id,
+        signin_date=today,
+        coin_awarded=rewards.get_signin_boost_coin(db),
+        ad_ref_id=ad_ref_id,
+    )
+    db.add(boost)
+    try:
+        acc, _ = crud_wallet.grant_coins(
+            db, user_id, boost.coin_awarded,
+            biz_type="signin_boost", ref_id=ad_ref_id or today.isoformat(),
+            remark=f"签到膨胀 第{record.cycle_day}天",
+        )
+        if commit:
+            db.commit()
+        else:
+            db.flush()
+    except IntegrityError as e:
+        db.rollback()
+        raise AlreadyBoostedError from e
+    db.refresh(boost)
+    return boost, acc.coin_balance

app/repositories/wallet.py

@@ -33,7 +33,6 @@ from app.models.wallet import (
 _WX_STATE_SUCCESS = "SUCCESS"
 _WX_STATE_FAILED = {"FAIL", "CANCELLED", "CLOSED"}
 _WX_STATE_WAIT_CONFIRM = "WAIT_USER_CONFIRM"  # 用户还没在微信确认页确认
-_WITHDRAW_ACTIVE_STATUSES = {"reviewing", "pending"}
 # 免确认收款授权状态
 _WX_AUTH_ACTIVE = "TAKING_EFFECT"   # 已生效,可免确认转账
 _WX_AUTH_CLOSED = "CLOSED"          # 已关闭(用户/商户/风控),需重新开启
@@ -63,10 +62,6 @@ class InsufficientCashError(Exception):
     """现金余额不足。"""
 
 
-class WithdrawTooFrequentError(Exception):
-    """提现申请过于频繁,或已有未完成提现单。"""
-
-
 class WithdrawTierUnavailableError(Exception):
     """该档位今日不可提:次数已满,或今天已选了其他额度(7-9 福利页档位规则)。"""
 
@@ -738,15 +733,6 @@ def create_withdraw(
     else:
         out_bill_no = uuid.uuid4().hex
 
-    active_order_id = db.execute(
-        select(WithdrawOrder.id).where(
-            WithdrawOrder.user_id == user_id,
-            WithdrawOrder.status.in_(_WITHDRAW_ACTIVE_STATUSES),
-        ).limit(1)
-    ).scalar_one_or_none()
-    if active_order_id is not None:
-        raise WithdrawTooFrequentError
-
     # 福利页档位闸(7-9):coin_cash 只能提预设档位,且该档今日可提(服务端权威口径,防绕过
     # 客户端刷)。放在幂等返回/在途互斥之后:同号重试仍原样返回旧单,不被档位闸误杀。
     # allow_sub_min(0.01 调试直发)保持原样放行,不受档位约束;invite_cash 本轮无档位概念不校验。
@@ -805,14 +791,6 @@ def create_withdraw(
         ).scalar_one_or_none()
         if existing is not None:
             return existing
-        active_order_id = db.execute(
-            select(WithdrawOrder.id).where(
-                WithdrawOrder.user_id == user_id,
-                WithdrawOrder.status.in_(_WITHDRAW_ACTIVE_STATUSES),
-            ).limit(1)
-        ).scalar_one_or_none()
-        if active_order_id is not None:
-            raise WithdrawTooFrequentError from None
         raise
     db.refresh(order)
     return order  # 待管理员审核;**不在此处打款**

app/schemas/ad.py

@@ -43,35 +43,6 @@ class AdRewardStatusOut(BaseModel):
     watch_seconds_remaining: int = Field(0, description="今日剩余可观看秒数;limit=0 时客户端不据此拦截")
 
 
-class AdRewardResultOut(BaseModel):
-    """按 ad_session_id 查本次广告的**权威发奖结果**(福利页看完视频的弹窗金额只认它)。
-
-    S2S 回调是异步的,客户端看完广告立刻轮询多半还查不到记录 —— 这种「还没到账」返回
-    200 + status='pending' 让客户端重试,**不返 404**:404 只应表示路由不存在,两者混在
-    一起客户端没法区分「后端没部署」和「再等等」。纯只读,不产生任何奖励。
-    """
-
-    ad_session_id: str = Field(..., description="回显请求的广告会话 id")
-    status: str = Field(
-        ...,
-        description="pending(S2S 未到账,客户端应继续轮询) / granted(已发奖) / capped(当日超限未发) / "
-        "ecpm_missing(缺 eCPM 未发) / closed_early(提前关闭未发);其余同 AdRewardRecord.status。"
-        "客户端只在 granted 且 coin>0 时弹窗,其它一律不弹(不显示假数字)",
-    )
-    coin: int | None = Field(
-        None,
-        description="本次实发金币:granted 为真实到账额;未发奖的状态为 0;pending 为 null",
-    )
-    round_coin: int | None = Field(
-        None,
-        description="**本轮膨胀累计已发金币**(含本条)——客户端「恭喜累计获得奖励」弹窗显示的就是它。"
-        "轮 = 用户点「去膨胀」到点「放弃赚钱」之间连看的若干条广告,边界由客户端的 boost_round_id 定。"
-        "本条不是 granted(capped/closed_early/…)时**仍返本轮累计**,只是这条按 0 计。"
-        "pending(没记录,取不到轮 id)、或该记录没有 boost_round_id(老客户端 / extra 丢失)时为 null,"
-        "客户端见 null 退回只显示单条 coin",
-    )
-
-
 class EcpmReportIn(BaseModel):
     """客户端上报一次广告展示的 eCPM(内部收益统计/对账)。
 
@@ -134,20 +105,13 @@ class TestGrantIn(BaseModel):
 
     reward_scene: str = Field(
         "reward_video",
-        description="模拟发奖场景。当前只支持 reward_video(普通激励视频);signin_boost(签到膨胀)"
-        "已于 2026-07 下线,传它会 422",
+        description="模拟发奖场景:reward_video(普通激励视频) / signin_boost(签到膨胀)",
     )
     ad_session_id: str | None = Field(
         None, min_length=8, max_length=64,
         description="本次广告会话 id(与 ecpm-report 同值)。reward_video 场景下据此查回客户端"
         "已上报的真实 eCPM 来按公式发奖;查不到或 eCPM≤0 时兜底 200,保证本地联调仍出非零金币",
     )
-    boost_round_id: str | None = Field(
-        None, max_length=64,
-        description="本次广告属于哪一轮膨胀。正式链路走穿山甲 S2S 的 mediaExtra,test-grant 不经 S2S、"
-        "拿不到 extra,故在 body 里补一个——不传的话 debug 包 reward-result 的 round_coin 恒为 null,"
-        "「弹窗 40 → 60 → toast +60」那套验收在本地跑不起来",
-    )
 
 
 class TestGrantOut(BaseModel):
@@ -155,7 +119,7 @@ class TestGrantOut(BaseModel):
 
     granted: bool = Field(..., description="本次是否真的发了金币(达每日上限则 False)")
     status: str = Field(
-        ..., description="granted / capped / ecpm_missing / unknown_scene"
+        ..., description="granted / capped / not_signed / already_boosted / last_day / unknown_scene"
     )
     coin: int = Field(..., description="本次发放金币(capped 时为 0)")
     used_today: int = Field(..., description="今日已成功发奖次数")

app/schemas/welfare.py

@@ -223,6 +223,16 @@ class SigninResultOut(BaseModel):
     coin_balance: int = Field(..., description="签到后金币余额")
 
 
+class SigninBoostRequest(BaseModel):
+    ad_ref_id: str | None = Field(None, description="广告会话/交易号。当前开发期可空,后续接 S2S 时回填")
+
+
+class SigninBoostResultOut(BaseModel):
+    coin_awarded: int = Field(..., description="本次膨胀补发金币")
+    coin_balance: int = Field(..., description="膨胀补发后金币余额")
+    signin_date: str = Field(..., description="被膨胀的签到日期 YYYY-MM-DD")
+
+
 # ===== 任务 =====
 
 class TaskOut(BaseModel):

app/services/llm_cost.py

@@ -1,53 +0,0 @@
-"""LLM 调用成本计算(纯逻辑,无 DB):按 model 分桶累加 token × 单价,返回总成本(元)+ 价格快照。
-
-用量取自 comparison_record.llm_calls[].usage(pricebot 已归一为 prompt/completion_tokens);
-error / 无 usage 的调用跳过。price_cfg = {per_model:{model:{input_per_1m,output_per_1m}}, default:{...}}。
-成本单位「元」——单次亚分级,用 float(不用 *_cents);snapshot 只含本次用到的模型的价(审计用,
-不存整张价表)。用到但没配价(既无 per_model 又无 default)的模型 → 快照标 unpriced,成本按 0 计。
-"""
-from __future__ import annotations
-
-_PRICE_KEY = "llm_token_price"
-
-
-def get_llm_prices(db) -> dict:
-    """读 LLM 单价配置(app_config;表内无则回退 CONFIG_DEFS 默认)。返回 compute_llm_cost 的 price_cfg。"""
-    from app.repositories import app_config  # 延迟 import:compute_llm_cost 纯逻辑不牵连 DB 层
-    return app_config.get_value(db, _PRICE_KEY)
-
-
-def compute_llm_cost(calls: list[dict], price_cfg: dict) -> tuple[float | None, dict | None]:
-    """遍历 calls 按 model 分桶,cost = Σ(入/1e6*入价 + 出/1e6*出价);无有效调用 → (None, None)。"""
-    if not calls:
-        return None, None
-    per_model = price_cfg.get("per_model") or {}
-    default = price_cfg.get("default")
-    buckets: dict[str, list[int]] = {}  # model -> [Σprompt_tokens, Σcompletion_tokens]
-    for c in calls:
-        if c.get("error"):
-            continue
-        usage = c.get("usage") or {}
-        model = c.get("model") or "unknown"
-        b = buckets.setdefault(model, [0, 0])
-        b[0] += usage.get("prompt_tokens") or 0
-        b[1] += usage.get("completion_tokens") or 0
-    if not buckets:  # 全是 error / 无 usage
-        return None, None
-    total = 0.0
-    prices: dict[str, dict] = {}
-    for model, (tin, tout) in buckets.items():
-        price = per_model.get(model, default)
-        in_p = price.get("input_per_1m") if isinstance(price, dict) else None
-        out_p = price.get("output_per_1m") if isinstance(price, dict) else None
-        # 没配价 / 无 default / 单价残缺或非法(配置页手改 JSON 可能存出脏数据)→ 标记待补价、
-        # 不计入成本;绝不抛异常,以免连累同一回填里的 token/llm_calls 落库。
-        if not isinstance(in_p, (int, float)) or not isinstance(out_p, (int, float)):
-            prices[model] = {"input_per_1m": in_p, "output_per_1m": out_p, "unpriced": True}
-            continue
-        total += tin / 1e6 * in_p + tout / 1e6 * out_p
-        prices[model] = {
-            "input_per_1m": in_p,
-            "output_per_1m": out_p,
-            "_source": "per_model" if model in per_model else "default",
-        }
-    return round(total, 6), {"mode": "per_model", "prices": prices}

docs/api/README.md

@@ -78,6 +78,7 @@
 | **签到**（前缀 `/api/v1/signin`） |||
 | 25 | `GET /api/v1/signin/status` | Bearer | [详情](./signin/signin-status.md) |
 | 26 | `POST /api/v1/signin` | Bearer | [详情](./signin/signin-do.md) |
+| 26a | `POST /api/v1/signin/boost` | Bearer | [详情](./signin/signin-boost.md) |
 | **任务**（前缀 `/api/v1/tasks`） |||
 | 27 | `GET /api/v1/tasks` | Bearer | [详情](./tasks/tasks-list.md) |
 | 28 | `POST /api/v1/tasks/{task_key}/claim` | Bearer | [详情](./tasks/tasks-claim.md) |
@@ -88,7 +89,6 @@
 | **看广告发奖**（前缀 `/api/v1/ad`） |||
 | 32 | `GET /api/v1/ad/pangle-callback` | 验签 | [详情](./ad/ad-pangle-callback.md) |
 | 33 | `GET /api/v1/ad/reward-status` | Bearer | [详情](./ad/ad-reward-status.md) |
-| 33a | `GET /api/v1/ad/reward-result/{ad_session_id}` | Bearer | [详情](./ad/ad-reward-result.md)（本次实发金币 + 本轮膨胀累计 `round_coin`,弹窗数字用它） |
 | 34 | `POST /api/v1/ad/test-grant` | Bearer | [详情](./ad/ad-test-grant.md) |
 | 35 | `POST /api/v1/ad/ecpm-report` | Bearer | [详情](./ad/ad-ecpm-report.md) |
 | 35a | `POST /api/v1/ad/feed-reward` | Bearer | [详情](./ad/ad-feed-reward.md) |

docs/api/ad/ad-pangle-callback.md

@@ -15,15 +15,7 @@ GroMore 以 GET 回调,关键参数:
 | `trans_id` | string | 交易号（**幂等键** + **唯一参与签名的字段**） |
 | `reward_name` | string | 奖励名（广告位配置,入库备注） |
 | `ecpm` | string\|null | GroMore 回调携带的 eCPM。普通激励视频优先用它计算金币 |
-| `extra` / `gromoreExtra` / `gromore_extra` | string | 客户端透传 JSON。支持 `ad_session_id`、`reward_scene`、`srv_env`、`boost_round_id` |
-
-### `extra` 里的 `boost_round_id`
-
-客户端生成的「这条广告属于哪一轮膨胀」标签(32 位十六进制,同 `ad_session_id` 格式),随发奖记录存进 `ad_reward_record.boost_round_id`。
-
-**它不参与任何发奖判定** —— 发多少、发不发完全不受影响,只是让 [`/ad/reward-result`](./ad-reward-result.md) 能把同一轮的 granted 记录求和成 `round_coin`(客户端「恭喜累计获得奖励」弹窗显示的数)。
-
-轮次边界由客户端定(只有它知道用户点没点「放弃赚钱」):点「去膨胀」新生成一个 → 点「继续看视频膨胀」复用同一个 → 点「放弃赚钱」/ ✕ / 返回 / 到每日上限 / 跨天 则丢弃。不带此字段(老客户端 / GroMore 偶发丢 extra)时存 NULL,`round_coin` 返 `null`。
+| `extra` / `gromoreExtra` / `gromore_extra` | string | 客户端透传 JSON。支持 `ad_session_id`、`reward_scene`;`reward_scene=signin_boost` 表示签到膨胀 |
 | `mediation_rit` | string | 代码位 ID（GroMore 带,目前仅入 raw 备查） |
 | `prime_rit` | string | 广告位 ID（同上） |
 | `adn_name` | string | 实际出广告的 ADN 名（同上,可用于收益分析） |
@@ -48,6 +40,6 @@ GroMore 以 GET 回调,关键参数:
 **发奖唯一可信入口**:验签 → 取 `user_id`/`extra` → 按 `reward_scene` 分流 → 幂等处理（按 `trans_id` 去重）。客户端不直接发奖,被破解也刷不到钱。
 
 - `reward_scene=reward_video` 或缺省:普通激励视频。金币按 `eCPM / 1000 * eCPM因子 * 当日次数因子 * 10000` 计算;若回调没有 `ecpm`,会按 `extra.ad_session_id` 查客户端 `/ad/ecpm-report` 的上报值;两边都没有 eCPM 时不发币,记录 `status=ecpm_missing`。
-- ~~`reward_scene=signin_boost`~~(签到膨胀):**2026-07 已下线**。它按固定 3000 金币发、与广告实际收益脱钩,产品确认非设计内口径。签到弹窗的「看广告膨胀」现与福利页看视频同走 `reward_video`。现在传 `signin_boost` 会落到「未知场景」分支(不发币,`status=unknown_scene`)。
+- `reward_scene=signin_boost`:签到膨胀。要求用户当天已签到且不是 Day14;看完视频固定发 `2000` 金币,写 `signin_boost_record` 与 `coin_transaction.biz_type=signin_boost`。
 - 未知 `reward_scene`:不发币,记录 `status=unknown_scene`,返回 `is_verify=false/reason=1`。
 - 验签过但参数缺/坏或 user 不存在 → 不发（`is_verify=false` + `reason`）;granted / capped / ecpm_missing / 业务不满足已记录 → `is_verify=true` + `reason=0`。

docs/api/ad/ad-reward-result.md

@@ -1,66 +0,0 @@
-# GET /api/v1/ad/reward-result/{ad_session_id} — 查本次广告的权威发奖结果 + 本轮累计
-
-客户端看完激励视频后轮询本接口，拿**本次实发金币**和**本轮累计**用于「恭喜累计获得奖励」弹窗。不再用余额差 / `coin_per_ad` 估算。
-
-**纯只读**：发奖仍只由验签过的 S2S 回调完成，本接口不写库、不产生任何奖励。按 `user_id` 收窄，被刷也只能查到自己的记录。
-
-## 鉴权
-
-需要 Bearer token。
-
-## 路径参数
-
-| 参数 | 类型 | 约束 | 说明 |
-|---|---|---:|---|
-| `ad_session_id` | string | 长度 8~64 | 本次广告会话 id，客户端生成，与 `mediaExtra` / `ecpm-report` 同值 |
-
-## 响应
-
-| 字段 | 类型 | 说明 |
-|---|---|---|
-| `ad_session_id` | string | 回显请求值 |
-| `status` | string | `pending`(S2S 未到账，继续轮询) / `granted` / `capped`(当日超限) / `ecpm_missing` / `closed_early`(提前关闭) |
-| `coin` | int \| null | **本条**实发金币。granted 为真实到账额；未发奖的状态为 0；pending 为 `null` |
-| `round_coin` | int \| null | **本轮累计已发金币**(含本条) ← 弹窗显示的就是它 |
-
-```json
-{ "ad_session_id": "3f2a9c1b7e4d8a60", "status": "granted", "coin": 20, "round_coin": 60 }
-```
-
-### `round_coin` 的口径
-
-「轮」= 用户点「去膨胀」到点「放弃赚钱」之间连看的若干条广告，边界由客户端的 `boost_round_id` 定（见 [ad-pangle-callback](./ad-pangle-callback.md)）。
-
-```sql
-SELECT COALESCE(SUM(coin), 0) FROM ad_reward_record
-WHERE user_id = :user_id                        -- 恒带，轮 id 是客户端生成的不可跨用户信任
-  AND boost_round_id = :该会话记录的 boost_round_id
-  AND status = 'granted'
-```
-
-由服务端求和而非客户端自己累加：客户端进程被杀 / 低内存重建后本地累计会丢，发奖记录不会。
-
-**要守住的不变量：弹窗数字 == 本轮实际到账之和 == 用户看到的余额涨幅。** 三者对不上，用户就会认为少发了钱。
-
-| 情形 | `round_coin` |
-|---|---|
-| 本条 `granted` | 本轮累计（含本条） |
-| 本条 `capped` / `closed_early` / `ecpm_missing` | **仍返本轮累计**，该条按 0 计（撞上限那下的 toast 要能显示前几条的总额，不能是空） |
-| `status=pending`（没记录） | `null` —— 连属于哪一轮都不知道。**不是 0**，0 会被读成「本轮没赚到」 |
-| 该记录没有 `boost_round_id`（老客户端 / extra 丢失） | `null`，客户端退回只显示单条 `coin` |
-
-## 错误
-
-- `401`: 未登录
-- `422`: `ad_session_id` 长度不在 8~64
-
-**查不到记录不返 404**，而是 200 + `status="pending"`。404 只应表示路由不存在；两者混在一起客户端没法区分「后端没部署」和「再等等」。
-
-## 实现注意
-
-同一 `ad_session_id` 可能有多条记录，取值时**显式优先 `granted`**，不能只取最近一条：
-
-- 客户端先报 `closed_early`、S2S 随后姗姗来迟 → 两条，`granted` 反而是后写的
-- 本地联调重复调 `test-grant` → 同 session 多条 `granted`（`trans_id` 各不相同）
-
-都没有 `granted` 才取最近一条，让客户端知道没发的原因。

docs/api/ad/ad-test-grant.md

@@ -9,8 +9,7 @@
 
 | 字段 | 类型 | 必填 | 默认 | 说明 |
 |---|---|---|---|---|
-| `reward_scene` | string | 否 | `reward_video` | 模拟发奖场景。当前**只支持** `reward_video`；`signin_boost`（签到膨胀）已于 2026-07 下线，传它返 `422` |
-| `boost_round_id` | string | 否 | `null` | 本次广告属于哪一轮膨胀。正式链路走 S2S 的 `mediaExtra`，本接口不经 S2S 拿不到 extra，故由 body 补。**不传的话 debug 包 `/ad/reward-result` 的 `round_coin` 恒为 `null`**，「弹窗 40 → 60 → toast +60」那套累计验收在本地跑不起来 |
+| `reward_scene` | string | 否 | `reward_video` | 模拟发奖场景。`reward_video`=普通激励视频；`signin_boost`=签到膨胀 |
 | `ad_session_id` | string(8~64) \| null | 否 | null | 本次广告会话 id(与 [ecpm-report](./ad-ecpm-report.md) 同值)。**仅 `reward_video` 场景生效**:据此查回客户端已上报的真实 eCPM,走与正式发奖相同的公式发奖;查不到或 eCPM≤0(测试应用常返 0/假值)时兜底 200,保证本地联调仍出非零金币 |
 
 ## 出参
@@ -34,4 +33,4 @@
 
 `reward_scene=reward_video` 时按上面 `ad_session_id` 查回的真实 eCPM 走金币公式发奖(取不到兜底 200)——便于本地用 [admin 金币审计](./admin-ad-coin-audit.md) 核对「看广告→金币」是否按公式计算。
 
-它让已登录客户端能自助发奖 = 绕过反作弊,**严禁在生产开启**。
+`reward_scene=signin_boost` 时复用签到膨胀业务规则:必须当天已签到、非第 14 天、当天未膨胀过,成功后写入 `signin_boost` 金币流水。它让已登录客户端能自助发奖 = 绕过反作弊,**严禁在生产开启**。

docs/api/admin/admin-stats-overview.md

@@ -37,8 +37,8 @@
 | `feed_ad_watch_count` | int | 信息流广告有效完成视频数(`ad_feed_reward_record.status=granted`) |
 | `signin_coin_total` | int | 签到累计发放金币(`biz_type=signin`) |
 | `signin_count` | int | 签到次数(`signin_record`) |
-| `signin_boost_coin_total` | int | **历史口径**:签到膨胀累计发放金币(`biz_type=signin_boost`)。功能已下线,数字不再增长,保留供对账 |
-| `signin_boost_watch_count` | int | **历史口径**:签到膨胀次数。膨胀 2026-07 已下线、`signin_boost_record` 表已 drop,改数 `coin_transaction.biz_type='signin_boost'` 的入账笔数(一次膨胀 = 一笔,与原口径等价),只会停在历史值不再增长 |
+| `signin_boost_coin_total` | int | 签到膨胀累计发放金币(`biz_type=signin_boost`) |
+| `signin_boost_watch_count` | int | 签到膨胀有效视频数(`signin_boost_record`) |
 
 **DashboardCash**
 | 字段 | 类型 | 说明 |

docs/api/signin/signin-boost.md

@@ -0,0 +1,33 @@
+# POST /api/v1/signin/boost — 签到后看广告膨胀金币
+
+用户 Day1-Day13 当天已签到后，看完一条激励视频，由穿山甲 S2S 回调固定补发 2000 金币。本接口只用于 S2S 发奖后的确认。
+
+## 鉴权
+
+需要 Bearer token。
+
+## 请求体
+
+| 字段 | 类型 | 必填 | 说明 |
+|---|---|---:|---|
+| `ad_ref_id` | string | 是 | 穿山甲 S2S 回调的 `trans_id`。回调需先以 `extra.reward_scene=signin_boost` 完成发奖 |
+
+## 响应
+
+| 字段 | 类型 | 说明 |
+|---|---|---|
+| `coin_awarded` | int | 本次膨胀补发金币 |
+| `coin_balance` | int | 补发后的金币余额 |
+| `signin_date` | string | 被膨胀的签到日期，格式 `YYYY-MM-DD` |
+
+## 错误
+
+- `401`: 未登录
+- `409`: 缺少/无效广告回调记录，非本人广告，回调未发奖，当天未签到，Day14，或当天已经膨胀过
+
+## 数据写入
+
+- 本接口不直接发奖;实际写入发生在 `/ad/pangle-callback` 的 `reward_scene=signin_boost` 分支。
+- 回调写 `signin_boost_record` 新增一行，用 `(user_id, signin_date)` 唯一约束防重复。
+- 回调使 `coin_account` 增加固定 `2000` 金币。
+- 回调写入 `coin_transaction.biz_type=signin_boost`。

docs/database/OVERVIEW.md

@@ -35,7 +35,7 @@
 | 资产卡 / 钱包余额 | [`coin_account`](./coin_account.md) | 一用户一行的金币+现金余额快照 |
 | 金币明细 | [`coin_transaction`](./coin_transaction.md) | 每次金币变动一笔流水 |
 | 现金明细 | [`cash_transaction`](./cash_transaction.md) | 每次现金变动一笔流水(分) |
-| 每日签到 | [`signin_record`](./signin_record.md) | 7 天循环发币。签到弹窗的「看广告膨胀」2026-07 起走 `reward_video`(按 eCPM 发,记 `ad_reward_record`),不再有独立的膨胀表 |
+| 每日签到 | [`signin_record`](./signin_record.md) + [`signin_boost_record`](./signin_boost_record.md) | 7 天循环发币;签到后看广告可膨胀一次 |
 | 一次性任务(开消息提醒等) | [`user_task`](./user_task.md) | 领一次发币 |
 | 看激励视频赚金币 | [`ad_reward_record`](./ad_reward_record.md) + [`ad_watch_log`](./ad_watch_log.md) + [`ad_ecpm_record`](./ad_ecpm_record.md) | 独立数据流:发奖 / 旧版观看时长 / 收益对账 |
 | 信息流/Draw 广告结算 | [`ad_feed_reward_record`](./ad_feed_reward_record.md) | 每展示满 10 秒累计一份奖励,完成后一次性入账;`ad_type`(feed/draw)+`feed_scene`(compare/coupon)分形态/场景 |
@@ -102,11 +102,12 @@
 | 注销 `DELETE /user` | `user` | U(软删:`phone→deleted_<id>`、`status=deleted`) |
 | 绑/解绑微信 `POST /wallet/bind-wechat`、`/unbind-wechat` | `user`.wechat_* | U |
 | 签到 `POST /signin/do` | `signin_record`(C) + `coin_account`(U) + `coin_transaction`(C `signin`) | 同事务 |
+| 签到膨胀 `POST /signin/boost` | `signin_boost_record`(C) + `coin_account`(U) + `coin_transaction`(C `signin_boost`) | 同事务;同日一次 |
 | 领任务 `POST /tasks/claim` | `user_task`(C) + `coin_account`(U) + `coin_transaction`(C `task_<key>`) | 同事务 |
 | 金币兑现金 `POST /wallet/exchange` | `coin_account`(U) + `coin_transaction`(C `exchange_out` −) + `cash_transaction`(C `exchange_in` +) | 同事务 |
 | 发起提现 `POST /wallet/withdraw` | `withdraw_order`(C `reviewing`,记 `source`) + `coin_account`(U 按 source 扣对应余额) + 流水(C −:`cash_transaction.withdraw` 或 `invite_cash_transaction.invite_withdraw`) | 同事务,**不打款**;#121 按 `source` 分账 |
 | 查提现状态 / 用户取消 `GET /wallet/withdraw/status` | `withdraw_order`(U) + 失败→对应账本退款流水(C `withdraw_refund` / `invite_withdraw_refund` +) | |
-| 穿山甲发奖 S2S 回调 `POST /ad/pangle-callback` | `ad_reward_record`(C)+ granted→`coin_account`(U)+`coin_transaction`(C `reward_video`) | `trans_id` 幂等 |
+| 穿山甲发奖 S2S 回调 `POST /ad/pangle-callback` | `ad_reward_record`(C)+ granted→`coin_account`(U)+`coin_transaction`(C `reward_video`/`signin_boost`) | `trans_id` 幂等 |
 | 看广告时长上报 `POST /ad/watch-report` | `ad_watch_log`(C) | |
 | 广告 eCPM 上报 `POST /ad/ecpm-report` | `ad_ecpm_record`(C) | |
 | 信息流广告结算 `POST /ad/feed-reward` | `ad_feed_reward_record`(C)+ granted→`coin_account`(U)+`coin_transaction`(C `feed_ad_reward`) | `client_event_id` 幂等 |
@@ -174,7 +175,7 @@
 ## 三、表间关系 & Join Key
 
 ### 硬外键(数据库 FK 约束)
-- **18 张用户维度表 `.user_id` → `user.id`**:`coin_account`(同时是 PK)、`coin_transaction`、`cash_transaction`、`invite_cash_transaction`、`withdraw_order`、`wechat_transfer_authorization`(同时是 PK)、`signin_record`、`user_task`、`comparison_record`(2026-07 起 `user_id` **可空**——harvest 帧0 建行时软鉴权可能拿不到)、`comparison_milestone_claim`、`savings_record`、`ad_reward_record`、`ad_watch_log`、`ad_ecpm_record`、`ad_feed_reward_record`、`price_report`、`feedback`、`device_liveness`。
+- **19 张用户维度表 `.user_id` → `user.id`**:`coin_account`(同时是 PK)、`coin_transaction`、`cash_transaction`、`invite_cash_transaction`、`withdraw_order`、`wechat_transfer_authorization`(同时是 PK)、`signin_record`、`signin_boost_record`、`user_task`、`comparison_record`(2026-07 起 `user_id` **可空**——harvest 帧0 建行时软鉴权可能拿不到)、`comparison_milestone_claim`、`savings_record`、`ad_reward_record`、`ad_watch_log`、`ad_ecpm_record`、`ad_feed_reward_record`、`price_report`、`feedback`、`device_liveness`。
 - `admin_audit_log.admin_id` → `admin_user.id`。
 - `price_report.comparison_record_id` → `comparison_record.id`(可空:关联记录被删后仍留上报历史)。
 - **邀请两表** → `user.id`:`invite_relation.inviter_user_id`、`invite_relation.invitee_user_id`(唯一)、`invite_fingerprint.inviter_user_id`——注意 FK 列名是 `inviter`/`invitee_user_id`,不是 `user_id`。
@@ -186,7 +187,7 @@
   | biz_type | ref_id 指向 | amount 符号 |
   |---|---|---|
   | `signin` | 当天日期串(= `signin_record.signin_date` 的 ISO `YYYY-MM-DD`) | + |
-  | `signin_boost`(**历史,2026-07 已下线**) | 当时的广告 `trans_id`,无则当天日期 ISO 串。不再产生新行,存量保留供对账 | + |
+  | `signin_boost` | 当天日期串(= `signin_boost_record.signin_date` 的 ISO `YYYY-MM-DD`) | + |
   | `task_<key>` | `user_task.task_key` | + |
   | `reward_video` / `ad_reward`(历史) | `ad_reward_record.trans_id` | + |
   | `feed_ad_reward` | `ad_feed_reward_record.client_event_id` | + |
@@ -224,7 +225,7 @@
 user ─1:1─ coin_account
 user ─1:1─ wechat_transfer_authorization
 user ─1:N─ { coin_transaction, cash_transaction, invite_cash_transaction, withdraw_order,
-            signin_record, user_task, comparison_record(user_id 可空),
+            signin_record, signin_boost_record, user_task, comparison_record(user_id 可空),
             comparison_milestone_claim, savings_record, ad_reward_record, ad_watch_log,
             ad_ecpm_record, ad_feed_reward_record, price_report, feedback, device_liveness }
                                              (device_liveness 硬 FK; (user_id,device_id) 唯一)
@@ -254,7 +255,7 @@ launch_confirm_sample                        (独立, 无硬 FK; 都上报不去
 
 1. **余额快照** `coin_account`:`coin_balance`(金币个数)+ `cash_balance_cents`(现金分)+ `invite_cash_balance_cents`(邀请奖励金分,#82),一用户一行,读取展示用。
 2. **流水账本** `coin_transaction` / `cash_transaction` / `invite_cash_transaction`:每次变动写一笔,`balance_after*` 记变动后余额,可逐笔回溯对账。**现金与邀请奖励金是两本物理隔离的账**——发放口径与提现对账各自独立。
-3. **唯一变动入口**:金币走 `repositories/wallet.grant_coins`,邀请奖励金走 `grant_invite_cash`——都是「更新快照 + 写流水,**不 commit**,由调用方同一事务 commit」。signin / task / ad_reward / feed_ad_reward / exchange / admin 走 `grant_coins`(`signin_boost` 2026-07 已下线,存量流水保留);`invite_reward` / admin 调整走 `grant_invite_cash`,靠 `biz_type` 区分来源。
+3. **唯一变动入口**:金币走 `repositories/wallet.grant_coins`,邀请奖励金走 `grant_invite_cash`——都是「更新快照 + 写流水,**不 commit**,由调用方同一事务 commit」。signin / signin_boost / task / ad_reward / feed_ad_reward / exchange / admin 走 `grant_coins`;`invite_reward` / admin 调整走 `grant_invite_cash`,靠 `biz_type` 区分来源。
 
 - **汇率**:`10000 金币 = 1 元 = 100 分`(`rewards.COIN_PER_YUAN`);兑换额必须是整分倍数。
 - **提现状态机**:`reviewing`(发起即原子扣款、待人工审核、**不打款**)→ 审核通过 `pending`(微信转账在途)→ `success` / `failed`(失败自动退款);审核拒绝 `rejected`(退款)。**按 `withdraw_order.source` 分账**(#121):`coin_cash` 单的扣款/退款写 `cash_transaction`,`invite_cash` 单写 `invite_cash_transaction`;`out_bill_no` 幂等,孤儿 pending 单由 `reconcile_pending_withdraws` 对账兜底,admin `withdraws/ledger-check` 分账校验「单 ↔ 流水」。

docs/database/README.md

@@ -35,6 +35,7 @@
 | `withdraw_order` | 提现单(现金→微信零钱,含人工审核态;`source` 分账 coin_cash/invite_cash) | `models/wallet.py` | [详情](./withdraw_order.md) |
 | `wechat_transfer_authorization` | 微信免确认转账授权(一用户一行) | `models/wallet.py` | [详情](./wechat_transfer_authorization.md) |
 | `signin_record` | 签到记录(7 天循环) | `models/signin.py` | [详情](./signin_record.md) |
+| `signin_boost_record` | 签到后看广告膨胀记录 | `models/signin.py` | [详情](./signin_boost_record.md) |
 | `user_task` | 一次性任务领取去重 | `models/task.py` | [详情](./user_task.md) |
 | `ad_reward_record` | 看激励视频发奖记录(S2S 回调,trans_id 幂等) | `models/ad_reward.py` | [详情](./ad_reward_record.md) |
 | `ad_watch_log` | 看广告观看时长(旧版兼容字段) | `models/ad_watch_log.py` | [详情](./ad_watch_log.md) |

docs/database/ad_reward_record.md

@@ -2,7 +2,7 @@
 
 > 模型 `app/models/ad_reward.py` · 仓库 `app/repositories/ad_reward.py` · 接口 [ad-pangle-callback](../api/ad-pangle-callback.md) / [ad-reward-status](../api/ad-reward-status.md) / [ad-test-grant](../api/ad-test-grant.md) · [← 索引](./README.md) · [总览](./OVERVIEW.md)
 
-每条 = 穿山甲一次**服务端激励回调**。`trans_id` 唯一做幂等键(穿山甲会重试,同号只处理一次)。`reward_scene` 区分普通激励视频、提现看视频等场景;`reward_date`(北京时间日期串)给普通激励视频"每日上限"计数用。
+每条 = 穿山甲一次**服务端激励回调**。`trans_id` 唯一做幂等键(穿山甲会重试,同号只处理一次)。`reward_scene` 区分普通激励视频、签到膨胀等场景;`reward_date`(北京时间日期串)给普通激励视频"每日上限"计数用。
 
 ## 用在哪 / 增删改查
 - **C(插入)**:`POST /ad/pangle-callback`(穿山甲 S2S,经 SHA256 验签;`grant_ad_reward` 或场景业务处理)或 `POST /ad/test-grant`(本地联调)。普通激励视频三道闸:① 验签不过 → API 层 403,不进库;② `trans_id` 已存在 → 原样返回不重复发;③ **当日发奖次数(`DAILY_AD_REWARD_LIMIT`,默认 500)到顶** → 记一行 `status='capped'`、`coin=0`、不发币。否则按 eCPM 公式发币。另:`POST /ad/reward-noshow`(`record_reward_noshow`,Bearer)在用户提前关/未发奖时记一行 `status='closed_early'`、`coin=0` 留痕(同 session 已 granted 则跳过)。
@@ -13,16 +13,15 @@
 | 列 | 类型 | 约束 / 默认 | 说明(取值 / join) |
 |---|---|---|---|
 | `id` | Integer | PK, autoincrement | |
-| `trans_id` | String(64) | UNIQUE, index, NOT NULL | 穿山甲交易号(幂等键)。**被 `coin_transaction.ref_id` 引用**(biz_type=reward_video 等)。`closed_early` 留痕记录无 S2S 交易号,用合成键 `noreward:{ad_session_id}` |
+| `trans_id` | String(64) | UNIQUE, index, NOT NULL | 穿山甲交易号(幂等键)。**被 `coin_transaction.ref_id` 引用**(biz_type=reward_video/signin_boost 等)。`closed_early` 留痕记录无 S2S 交易号,用合成键 `noreward:{ad_session_id}` |
 | `user_id` | Integer | FK→user.id, index, NOT NULL | 归属用户(回调 media_extra 带回;不存在抛 UnknownUserError) |
-| `reward_scene` | String(32) | NOT NULL, default `reward_video` | 奖励场景:`reward_video` 普通激励视频(当前唯一发币场景);`withdrawal_ad` 提现门槛视频(只留痕不发币);`signin_boost` **历史值,2026-07 已下线** |
+| `reward_scene` | String(32) | NOT NULL, default `reward_video` | 奖励场景:`reward_video` 普通激励视频;`signin_boost` 签到膨胀 |
 | `ad_session_id` | String(64) | index, nullable | 客户端广告会话 ID,来自 `extra.ad_session_id`;用于匹配 `ad_ecpm_record` |
-| `boost_round_id` | String(64) | nullable | 「这条广告属于哪一轮膨胀」,来自 `extra.boost_round_id`。一轮 = 用户点「去膨胀」到点「放弃赚钱」之间连看的若干条。**纯标签,不参与发奖判定**;仅供 `/ad/reward-result` 求和出 `round_coin`(弹窗显示的累计值)。老客户端 / extra 丢失时 NULL |
 | `ecpm_raw` | String(32) | nullable | 本次发奖采用的 eCPM 原始值;可来自 S2S `ecpm` 或客户端上报 |
 | `app_env` | String(16) | nullable | 来源应用 `prod`(傻瓜比价)/`test`(测试);S2S 不带,发奖时按 `ad_session_id` 匹配 `ad_ecpm_record` 回填,查不到 NULL。广告收益报表金币侧按它聚合 |
 | `our_code_id` | String(64) | nullable | 我们配置的代码位 104xxx(同上回填) |
 | `coin` | Integer | NOT NULL, default 0 | 实发金币;`capped`/`ecpm_missing`/`closed_early`/业务不满足时为 0 |
-| `status` | String(16) | NOT NULL, default `granted` | 取值:`granted`(已发)/ `capped`(当日次数超限)/ `ecpm_missing`(缺 eCPM)/ `closed_early`(展示了但用户提前关/跳过,未发奖,客户端 reward-noshow 留痕)/ `unknown_scene`(回调 `reward_scene` 不在支持集合,只留痕不发) |
+| `status` | String(16) | NOT NULL, default `granted` | 取值:`granted`(已发)/ `capped`(当日次数超限)/ `ecpm_missing`(缺 eCPM)/ `closed_early`(展示了但用户提前关/跳过,未发奖,客户端 reward-noshow 留痕)/ `not_signed`/`already_boosted`/`last_day`/ `unknown_scene`(回调 `reward_scene` 不在支持集合,只留痕不发) |
 | `reward_date` | String(10) | index, NOT NULL | 北京时间日期串 `YYYY-MM-DD`,按它等值统计当日发奖次数 |
 | `reward_name` | String(64) | nullable | 穿山甲上报奖励名(参考,不作发奖依据) |
 | `raw` | String(1024) | nullable | 回调原始参数(审计排查) |
@@ -35,10 +34,8 @@
 
 ## 索引与约束
 - PK `id`;UNIQUE+index `trans_id`;index `user_id`、`reward_date`、`created_at`、`ad_session_id`。
-- 复合 index `ix_ad_reward_user_boost_round` = `(user_id, boost_round_id)`:算「本轮累计已发」用。**求和恒带 `user_id`** —— `boost_round_id` 是客户端生成的,不带 `user_id` 等于让任何人拿别人的轮 id 查别人发了多少。
 
 ## 注意
 - 普通激励视频按 eCPM 公式发奖;若 S2S 与客户端会话上报都缺 eCPM,记录 `status='ecpm_missing'`、`coin=0`,不发币。
-- 签到膨胀(`reward_scene=signin_boost`)2026-07 已下线,存量行保留供对账;签到弹窗的「看广告膨胀」现与福利页看视频同走 `reward_video`(按 eCPM 公式发)。
-- **膨胀轮累计**:`SUM(coin) WHERE user_id=? AND boost_round_id=? AND status='granted'`,由 `/ad/reward-result` 返回为 `round_coin`。客户端就算一直复用同一个轮 id,也只是把展示数字滚大 —— 求和的是**已发生**的发奖记录,不产生任何新入账,无资损风险。
+- 签到膨胀复用本表记录 S2S 幂等,实发固定 `2000` 金币由 `signin_boost_record`/`coin_transaction.biz_type=signin_boost` 承载。
 - 并发同 `trans_id` 撞唯一约束 → catch IntegrityError 回滚返回已存在那条(幂等兜底)。

docs/database/app_config.md

@@ -14,7 +14,7 @@
 ## 字段
 | 列 | 类型 | 约束 / 默认 | 说明(取值 / join) |
 |---|---|---|---|
-| `key` | String(64) | **PK** | 配置标识,取值见 `config_schema.CONFIG_DEFS`:`signin_rewards` / `min_exchange_coin` / `withdraw_min_cents` / `withdraw_max_cents` / `task_rewards` / `record_milestones` / `ad_reward_coin` / `ad_daily_limit` / `ad_max_coin` / `ad_round_count` / `ad_cooldown_sec` / `withdraw_auto_reconcile_enabled` / `comparing_ad_enabled` |
+| `key` | String(64) | **PK** | 配置标识,取值见 `config_schema.CONFIG_DEFS`:`signin_rewards` / `min_exchange_coin` / `withdraw_min_cents` / `withdraw_max_cents` / `task_rewards` / `record_milestones` / `ad_reward_coin` / `ad_daily_limit` / `ad_max_coin` / `ad_round_count` / `ad_cooldown_sec` / `signin_boost_coin` / `withdraw_auto_reconcile_enabled` / `comparing_ad_enabled` |
 | `value` | JSON(PG: JSONB) | NOT NULL | 配置值,类型随 key(`int` / `int_list` 如签到 14 档 / `dict_str_int` 如 task_rewards / `bool` 如 withdraw_auto_reconcile_enabled / comparing_ad_enabled) |
 | `updated_by_admin_id` | Integer | nullable | 最后修改的管理员 id(= `admin_user.id`,软引用,无 FK) |
 | `updated_at` | DateTime(tz) | server_default now(), onupdate now() | 最后修改时间 |

docs/database/coin_transaction.md

@@ -10,7 +10,7 @@
   | 动作 / endpoint | `biz_type` | `amount` | `ref_id` 指向 |
   |---|---|---|---|
   | 签到 `POST /signin/do` | `signin` | + | 当天日期串(= `signin_record.signin_date` ISO) |
-  | ~~签到后看广告膨胀~~(**2026-07 已下线**) | `signin_boost` | + | 历史行:当时的广告 `trans_id`,无则当天日期 ISO 串。不再产生新行;签到弹窗的看广告改走 `reward_video` |
+  | 签到后看广告膨胀 `POST /signin/boost` | `signin_boost` | + | 广告 `trans_id`(= `signin_boost_record.ad_ref_id`);无 ad_ref_id 时回退当天日期 ISO 串 |
   | 领任务 `POST /tasks/claim` | `task_<key>`(如 `task_enable_notification`) | + | 一次性任务=`user_task.task_key`;可重复任务(`enable_notification`)=带序号 `task_key:N` |
   | 普通激励视频 S2S 回调 `POST /ad/pangle-callback` | `reward_video`(历史兼容:`ad_reward`) | + | `ad_reward_record.trans_id` |
   | 信息流广告结算 `POST /ad/feed-reward` | `feed_ad_reward` | + | `ad_feed_reward_record.client_event_id` |
@@ -36,7 +36,7 @@
 
 ## 关系 / Join Key
 - `user_id` → `user.id`(多对一)。
-- `ref_id` 是**软关联**(无 FK),目标随 `biz_type`:`signin`→签到日(`signin_record.signin_date` ISO) / `signin_boost`(历史)→当时的广告 `trans_id`(无则当天日期) / `task_<key>`→一次性任务=`user_task.task_key`、可重复任务=`task_key:N` / `reward_video`/`ad_reward`→`ad_reward_record.trans_id` / `feed_ad_reward`→`ad_feed_reward_record.client_event_id` / 其余 null。
+- `ref_id` 是**软关联**(无 FK),目标随 `biz_type`:`signin`→签到日(`signin_record.signin_date` ISO) / `signin_boost`→`signin_boost_record.ad_ref_id`(无则当天日期) / `task_<key>`→一次性任务=`user_task.task_key`、可重复任务=`task_key:N` / `reward_video`/`ad_reward`→`ad_reward_record.trans_id` / `feed_ad_reward`→`ad_feed_reward_record.client_event_id` / 其余 null。
 
 ## 索引与约束
 - PK `id`;index `user_id`、`created_at`。

docs/database/comparison_record.md

@@ -40,8 +40,6 @@
 | `raw_payload` | JSON(PG: JSONB) | nullable | 客户端原始上报全量(calibration + done.params),取数兜底 |
 | `input_tokens` | Integer | nullable | 本次 LLM 累计输入 token = Σ `llm_calls[].usage.prompt_tokens`(server 收上报后从 `llm_calls` 累加;旧记录/未采集为 null) |
 | `output_tokens` | Integer | nullable | 本次 LLM 累计输出 token = Σ `llm_calls[].usage.completion_tokens`(同上) |
-| `llm_cost_yuan` | Float | nullable | 本次比价 LLM 总成本(元),回填时按「当时价」逐模型算好冻结(见 `services/llm_cost.py`);旧记录/未回填为 null → 前端回退「估算成本」 |
-| `llm_price_snapshot` | JSON(PG: JSONB) | nullable | 算成本所用单价快照 `{mode, prices:{model:{input_per_1m,output_per_1m,_source}}}`;`app_config` 只存当前价、不留历史,故冻结当时价供审计/复算 |
 | `created_at` | DateTime(tz) | server_default now(), index | 时间 |
 
 > `ordered`(已下单)是**瞬态字段**,不在表里:`list_records` 读取时按 `store_name ∈ 该用户 source='compare' 的 savings_record.shop_name 集合` 现挂到实例上供出参用。

docs/database/signin_boost_record.md

@@ -0,0 +1,22 @@
+# signin_boost_record — 签到膨胀记录
+
+App 用户当天签到后，看完激励视频可固定膨胀一次（默认 3000 金币，`rewards.SIGNIN_BOOST_COIN`，运营后台 `app_config.signin_boost_coin` 可改）。循环最后一天（`cycle_day == SIGNIN_CYCLE_LEN`，即 7 天循环的第 7 天）不展示也不允许膨胀。本表记录膨胀动作，并用唯一约束防重复补发。
+
+## 字段
+
+| 字段 | 类型 | 约束 | 说明 |
+|---|---|---|---|
+| `id` | Integer | PK | 自增主键 |
+| `user_id` | Integer | FK → `user.id`, index, NOT NULL | 用户 |
+| `signin_date` | Date | NOT NULL | 被膨胀的签到日期，北京时间 |
+| `coin_awarded` | Integer | NOT NULL | 本次补发金币,默认固定 3000(`rewards.get_signin_boost_coin`) |
+| `ad_ref_id` | String(64) | nullable | 穿山甲 S2S 回调 `trans_id` |
+| `created_at` | DateTime(tz) | server_default now(), NOT NULL | 创建时间 |
+
+## 约束
+
+- `UNIQUE(user_id, signin_date)` = `uq_signin_boost_user_date`：同一用户同一天只能膨胀一次。
+
+## 关联
+
+- 膨胀成功时写 `coin_transaction.biz_type=signin_boost`，`ref_id = ad_ref_id`(无 ad_ref_id 时回退当天日期 ISO 串)。

docs/database/user.md

@@ -29,7 +29,7 @@
 | `last_login_at` | DateTime(tz) | 应用层 default utcnow | 最近登录时间(每次登录更新) |
 
 ## 关系 / Join Key
-- **被引用方(本表是 1,对方是 N/1)**:`coin_account`、`coin_transaction`、`cash_transaction`、`withdraw_order`、`wechat_transfer_authorization`、`signin_record`、`user_task`、`comparison_record`、`comparison_milestone_claim`、`savings_record`、`ad_reward_record`、`ad_watch_log`、`ad_ecpm_record`、`ad_feed_reward_record`、`price_report`、`feedback` 的 `user_id` 均 → `user.id`;`invite_relation` 的 `inviter_user_id` / `invitee_user_id` 均 → `user.id`。
+- **被引用方(本表是 1,对方是 N/1)**:`coin_account`、`coin_transaction`、`cash_transaction`、`withdraw_order`、`wechat_transfer_authorization`、`signin_record`、`signin_boost_record`、`user_task`、`comparison_record`、`comparison_milestone_claim`、`savings_record`、`ad_reward_record`、`ad_watch_log`、`ad_ecpm_record`、`ad_feed_reward_record`、`price_report`、`feedback` 的 `user_id` 均 → `user.id`;`invite_relation` 的 `inviter_user_id` / `invitee_user_id` 均 → `user.id`。
 - 与 `admin_user` **无任何关联**(C 端用户 vs 后台管理员,两套体系)。
 
 ## 索引与约束

docs/database/withdraw_order.md

@@ -45,9 +45,9 @@ reviewing ──admin 审核拒绝──▶ rejected(已退款)
 
 ## 索引与约束
 - PK `id`;UNIQUE+index `out_bill_no`;index `user_id`、`created_at`。
-- 部分唯一索引 `ux_withdraw_order_user_active`(`user_id`),条件 `status IN ('reviewing', 'pending')`:每个用户同时只能有一笔在途(待审核 / 打款中)提现单,DB 层挡并发重复提现。
+- 同一用户允许同时存在多笔 `reviewing` / `pending` 提现单;每笔以唯一 `out_bill_no` 独立审核、打款和对账。
 
 ## 注意
-- **资金安全**:原子扣款(`WHERE cash_balance_cents >= amount`)+ `out_bill_no` 幂等 + 结果不明时**先查单再决定,绝不盲目退款**(防退款后又到账)+ 孤儿 pending 单 `reconcile_pending_withdraws` 对账兜底。
+- **资金安全**:允许多笔在途不等于重复扣款。仍由原子扣款(`WHERE cash_balance_cents >= amount`)+ `out_bill_no` 幂等 + 结果不明时**先查单再决定,绝不盲目退款**(防退款后又到账)+ 孤儿 pending 单 `reconcile_pending_withdraws` 对账兜底。
 - `WITHDRAW_MIN_CENTS=10`(0.1 元,微信商家转账地板价),可经 `app_config.withdraw_min_cents` 后台调。
 - "待审核期间钱已扣减",防用户拿同一笔余额重复发起多笔提现。

docs/database/数据表字典-产品参考.md

@@ -27,6 +27,7 @@
 | 钱包 | `withdraw_order` | 提现单 | 现金→微信零钱提现 |
 | 钱包 | `wechat_transfer_authorization` | 微信转账免确认授权 | 用户授权后转账免逐笔确认 |
 | 激励 | `signin_record` | 签到记录 | 每日签到 |
+| 激励 | `signin_boost_record` | 签到膨胀记录 | 签到后看广告翻倍补发 |
 | 激励 | `user_task` | 一次性任务完成 | 只能领一次的任务 |
 | 激励 | `comparison_milestone_claim` | 比价战绩领取 | 比价次数里程碑奖励 |
 | 比价 | `comparison_record` | 比价记录 | 用户视角「我的比价记录」 |
@@ -133,7 +134,7 @@ App 用户主表。两种登录（极光一键 / 短信验证码）都映射到
 
 ## `withdraw_order` — 提现单
 
-现金 → 微信零钱。状态机：reviewing（待审核）→ pending（打款在途）→ success / failed；reviewing →（拒绝）→ rejected（已退款）。同一用户同时只能有一笔进行中。
+现金 → 微信零钱。状态机：reviewing（待审核）→ pending（打款在途）→ success / failed；reviewing →（拒绝）→ rejected（已退款）。同一用户可同时存在多笔进行中的提现单，每次提交通过唯一 `out_bill_no` 保证幂等。
 
 | 字段 | 类型 | 说明 |
 |---|---|---|
@@ -180,6 +181,18 @@ App 用户主表。两种登录（极光一键 / 短信验证码）都映射到
 | coin_awarded | 整数 | 本次发放金币 |
 | created_at | 时间 | 时间 |
 
+## `signin_boost_record` — 签到膨胀记录
+
+签到后看广告「膨胀」翻倍，一天最多一次，补发金额=当天签到原始奖励。
+
+| 字段 | 类型 | 说明 |
+|---|---|---|
+| id | 整数 | **主键** |
+| user_id | 整数 | **外键→user** |
+| signin_date | 日期 | 与 user_id 组成**唯一**（防并发重复补发） |
+| coin_awarded | 整数 | 补发金币 |
+| ad_ref_id | 字符串 | 广告会话/交易号，可空 |
+| created_at | 时间 | 时间 |
 
 ## `user_task` — 一次性任务完成记录
 
@@ -366,7 +379,7 @@ profile「累计帮你省了」「省钱战绩」的唯一数据源。(user_id,
 | user_id | 整数 | **外键→user** |
 | coin | 整数 | 实发金币（超限为 0） |
 | status | 字符串 | granted（已发）/ capped（当日超限）/ ecpm_missing（缺 eCPM） |
-| reward_scene | 字符串 | reward_video（福利页看视频，当前唯一发币场景）/ withdrawal_ad（提现门槛视频，不发币）/ signin_boost（**历史值，2026-07 已下线**） |
+| reward_scene | 字符串 | reward_video（福利页看视频）/ signin_boost（签到膨胀） |
 | ad_session_id | 字符串 | 广告会话 id，可空 |
 | ecpm_raw | 字符串 | 本次发奖采用的 eCPM 原始值，可空 |
 | app_env | 字符串 | 应用环境 prod/test（回填），可空 |

docs/guides/看广告赚金币上线清单.md

@@ -70,7 +70,7 @@
 ## C. 部署 + 包名
 
 - [ ] **后端部署到公网**(由服务器管理员;`/opt/shaguabijia-app-server`,uvicorn 127.0.0.1:8770,nginx 反代)
-- [ ] **跑迁移**:`alembic upgrade head`(包含 `ad_reward_record`、`ad_feed_reward_record` 等表)
+- [ ] **跑迁移**:`alembic upgrade head`(包含 `ad_reward_record`、`signin_boost_record`、`ad_feed_reward_record` 等表)
 - [ ] **包名定稿**:当前 `com.jishisongfu.shaguabijia`。穿山甲(APP_ID 5830519)、极光、微信都绑"包名 + 签名",定了再上,别再换
   - 微信提现链路当前因复用 elderhelper 的 appid + 包名切换已 dead,要恢复需申请傻瓜比价自己的微信 appid(另见客户端 build.gradle 注释)
 - [ ] (可选,提升真实填充)集成 **MSA OAID SDK**:申请证书(绑包名、审核几天)。App 侧当前 `getDevOaid=null`,有 OAID 后投放匹配 + 填充会明显改善

scripts/reset_onboarding.py

@@ -1,138 +0,0 @@
-"""重置指定用户的新手引导完成标记,让这个账号重新进新手引导页,方便反复测试引导流程。
-
-原理:是否跳过引导只由 onboarding_completion 表里 (user_id, device_id) 那一行决定
-(见 app/models/onboarding.py)。删掉该用户的行 → 登录响应 onboarding_completed=false、
-GET /api/v1/user/onboarding/status 也返 false → 客户端下次登录/启动重走引导。
-本地 SharedPreferences 标记卸载即丢、以后端为准,所以删这一行就够,不用重装 App。
-
-默认删该用户**所有设备**的记录(换机/多设备一起放开);只想放开某一台用 --device-id
-(device_id = 客户端硬件级 ANDROID_ID,与登录 / onboarding/complete 传的是同一个值)。
-
-与已有两个入口的分工:
-  - admin「设备维度引导管理」按**设备**重置(该设备上所有账号一起),本脚本按**账号**;
-  - POST /api/v1/user/onboarding/reset 要客户端自己带 device_id 调,本脚本从库里反查设备。
-
-用法(在项目根、已 pip install -e . 的环境里跑):
-  python scripts/reset_onboarding.py                     # 默认测试号 11111111111
-  python scripts/reset_onboarding.py 13800138000         # 指定手机号
-  python scripts/reset_onboarding.py --user-id 5         # 直接指定 user_id
-  python scripts/reset_onboarding.py --dry-run           # 预览(照常执行再回滚),不落库
-  python scripts/reset_onboarding.py --device-id abc123  # 只放开这一台设备,其余设备照旧跳过
-
-走 SessionLocal 连 DATABASE_URL(SQLite / Postgres 都行),因此**默认只允许 APP_ENV=dev 改库**
-(--dry-run 只读,任何环境都能跑)。线上确实要给某个用户开引导时加 --force —— 这张表只存
-"引导走过没"的标记,删了最坏结果是用户多看一次引导,不涉及金额/账目。
-"""
-from __future__ import annotations
-
-import argparse
-import sys
-
-from sqlalchemy import delete, select
-
-from app.core.config import settings
-from app.db.session import SessionLocal, engine
-from app.models.onboarding import OnboardingCompletion
-from app.models.user import User
-
-# Windows 控制台默认 GBK,强制 UTF-8 否则中文输出乱码。stderr 也要设:
-# SystemExit(如"用户不存在")的中文提示走的是 stderr。
-for _stream in (sys.stdout, sys.stderr):
-    if hasattr(_stream, "reconfigure"):
-        _stream.reconfigure(encoding="utf-8")
-
-# dev 下 engine 是 echo=True(APP_DEBUG),几十行 SQL 会把前后对比刷没。echo 走 SQLAlchemy 自己的
-# InstanceLogger,不吃 logging.setLevel,只能改 engine.echo。
-engine.echo = False
-
-DEFAULT_PHONE = "11111111111"
-
-
-def resolve_user(db, phone: str, user_id: int | None) -> User:
-    if user_id is not None:
-        user = db.get(User, user_id)
-        if user is None:
-            raise SystemExit(f"user_id={user_id} 不存在")
-        return user
-    user = db.execute(select(User).where(User.phone == phone)).scalar_one_or_none()
-    if user is None:
-        raise SystemExit(f"手机号 {phone} 没有对应用户(注意 phone 才是登录账号,username 是展示 ID)")
-    return user
-
-
-def print_state(db, user: User, device_id: str | None, label: str) -> None:
-    """打印该用户当前的引导完成标记。--device-id 时只看那一台,便于确认没误伤别的设备。"""
-    stmt = (
-        select(OnboardingCompletion.device_id, OnboardingCompletion.completed_at)
-        .where(OnboardingCompletion.user_id == user.id)
-        .order_by(OnboardingCompletion.completed_at.desc())
-    )
-    if device_id:
-        stmt = stmt.where(OnboardingCompletion.device_id == device_id)
-    rows = db.execute(stmt).all()
-
-    print(f"--- {label} ---")
-    scope = f"device_id={device_id}" if device_id else "全部设备"
-    if not rows:
-        print(f"  onboarding_completion({scope}): (无) → 该用户会走引导")
-        return
-    print(f"  onboarding_completion({scope}): {len(rows)} 条 → 这些设备上会跳过引导")
-    for did, at in rows:
-        print(f"    device_id={did}  完成于 {at}")
-
-
-def main() -> None:
-    parser = argparse.ArgumentParser(description="重置指定用户的新手引导,让其重新进引导页")
-    parser.add_argument("phone", nargs="?", default=DEFAULT_PHONE,
-                        help=f"手机号(默认 {DEFAULT_PHONE})")
-    parser.add_argument("--user-id", type=int, default=None, help="直接按 user_id 定位,优先于 phone")
-    parser.add_argument("--device-id", default=None,
-                        help="只重置这一台设备(硬件级 ANDROID_ID);默认重置该用户所有设备")
-    parser.add_argument("--dry-run", action="store_true", help="预览,最后回滚不落库")
-    parser.add_argument("--force", action="store_true", help="非 dev 环境也允许改库(仅删引导标记,不涉及账目)")
-    args = parser.parse_args()
-
-    if not args.dry_run and settings.APP_ENV != "dev" and not args.force:
-        raise SystemExit(
-            f"APP_ENV={settings.APP_ENV},默认只有 dev 能改库。确认要在该环境重置请加 --force"
-            "(--dry-run 只读,任意环境可跑)"
-        )
-
-    db = SessionLocal()
-    try:
-        user = resolve_user(db, args.phone, args.user_id)
-        print(f"DB: {settings.DATABASE_URL}  APP_ENV: {settings.APP_ENV}")
-        print(f"用户: id={user.id} phone={user.phone} username={user.username}")
-        print_state(db, user, args.device_id, "before")
-
-        stmt = delete(OnboardingCompletion).where(OnboardingCompletion.user_id == user.id)
-        if args.device_id:
-            stmt = stmt.where(OnboardingCompletion.device_id == args.device_id)
-        deleted = db.execute(stmt).rowcount
-
-        if not deleted:
-            # 没记录本来就会走引导 —— 常见于:换了新设备、或运营/客户端已经重置过一次。
-            print("该用户(该范围内)本来就没有完成标记,已经会走引导了,无需处理。")
-            db.rollback()
-        else:
-            print_state(db, user, args.device_id, "after")
-            if args.dry_run:
-                db.rollback()
-                print(f"(dry-run:以上 after 为预览,已回滚,库没动;真跑会删 {deleted} 条)")
-                return
-            db.commit()
-            print(f"完成:删掉 {deleted} 条完成标记,{user.phone} 下次登录会重走新手引导。")
-
-        # 测试号无论库里有没有记录都恒走引导(见 app/core/test_account.py),提醒一句免得白跑
-        if settings.test_account_phone and user.phone == settings.test_account_phone:
-            print(f"提示:{user.phone} 是配置的测试账号(TEST_ACCOUNT_PHONE),"
-                  "登录响应 onboarding_completed 恒为 false、本就每次都走引导,无需重置。")
-
-        print("提醒:客户端是在登录响应 / 启动时查 onboarding/status 的,已经在首页的 App 不会自动跳转,"
-              "退出登录重进(或杀掉重开)才会看到引导页。")
-    finally:
-        db.close()
-
-
-if __name__ == "__main__":
-    main()

scripts/reset_signin_today.py

@@ -1,218 +0,0 @@
-"""撤销指定用户「今天已签到」的状态,让今天可以重新签到,方便反复测试签到流程。
-
-与 reset_signin.py 的区别:那个删**全部**签到历史(连续天数从头再来);本脚本只精确撤销
-**今天**这一次,昨天及以前的记录原样保留 —— 所以重签后 cycle_day / streak 会接着昨天继续,
-7 天循环的档位不会被打乱,可以连着好几天测「第 N 档」的奖励。
-
-默认是**完整撤销**(等于今天这次签到从没发生过):
-  1. 删 signin_record 今天这行 → 今天变回未签到
-  2. 删今天的 signin 金币流水,并把金币从 coin_account 余额 / 累计收益里扣回
-
-金币默认要退:签到流水**没有**唯一索引拦重复(ux_coin_transaction_task_ref 只覆盖
-biz_type LIKE 'task%'),不退的话每测一轮余额就白涨一次奖励,coin_transaction 里还会堆出
-同一 ref_id(日期)的重复流水,收益明细页会看到两条今天的签到。真想留着奖励用 --keep-coins。
-
-例外:签到的金币若已被兑换成现金(余额已不够退),**自动跳过退款**并保留今天的签到流水。
-因为 coin_balance 必须恒等于流水总和,硬退会把余额退成负数 —— 夹到 0 又会吃掉别处赚的金币,
-两种做法都会让账对不上。这时重签会再发一次奖励,余额多涨一档,属可接受的测试噪音。
-
-用法(在项目根、已 pip install -e . 的环境里跑):
-  python scripts/reset_signin_today.py                 # 默认测试号 11111111111
-  python scripts/reset_signin_today.py 13800138000     # 指定手机号
-  python scripts/reset_signin_today.py --user-id 5     # 直接指定 user_id
-  python scripts/reset_signin_today.py --dry-run       # 预览(照常执行再回滚),不落库
-  python scripts/reset_signin_today.py --keep-coins    # 只删签到记录,保留已发金币
-
-「今天」直接复用 app.core.rewards.cn_today(北京时间),与签到判重同源,不自己算时区。
-走 SessionLocal 连 DATABASE_URL(SQLite / Postgres 都行),因此**只允许 APP_ENV=dev 时改库**
-(--dry-run 只读,任何环境都能跑)。
-"""
-from __future__ import annotations
-
-import argparse
-import sys
-
-from sqlalchemy import select
-
-from app.core.config import settings
-from app.core.rewards import cn_today
-from app.db.session import SessionLocal, engine
-from app.models.signin import SigninRecord
-from app.models.user import User
-from app.models.wallet import CoinAccount, CoinTransaction
-from app.repositories import signin as crud_signin
-
-# Windows 控制台默认 GBK,强制 UTF-8 否则中文输出乱码
-if hasattr(sys.stdout, "reconfigure"):
-    sys.stdout.reconfigure(encoding="utf-8")
-
-# dev 下 engine 是 echo=True(APP_DEBUG),几十行 SQL 会把前后对比刷没。echo 走 SQLAlchemy 自己的
-# InstanceLogger,不吃 logging.setLevel,只能改 engine.echo。
-engine.echo = False
-
-DEFAULT_PHONE = "11111111111"
-
-
-def resolve_user(db, phone: str, user_id: int | None) -> User:
-    if user_id is not None:
-        user = db.get(User, user_id)
-        if user is None:
-            raise SystemExit(f"user_id={user_id} 不存在")
-        return user
-    user = db.execute(select(User).where(User.phone == phone)).scalar_one_or_none()
-    if user is None:
-        raise SystemExit(f"手机号 {phone} 没有对应用户(注意 phone 才是登录账号,username 是展示 ID)")
-    return user
-
-
-
-def print_state(db, user: User, today, label: str) -> None:
-    print(f"--- {label} ---")
-    rec = db.execute(
-        select(SigninRecord).where(
-            SigninRecord.user_id == user.id, SigninRecord.signin_date == today
-        )
-    ).scalar_one_or_none()
-    if rec is None:
-        print(f"  signin_record {today}: (无)")
-    else:
-        print(f"  signin_record {today}: 第{rec.cycle_day}档 连续{rec.streak}天 +{rec.coin_awarded}金币")
-
-    last = db.execute(
-        select(SigninRecord.signin_date)
-        .where(SigninRecord.user_id == user.id)
-        .order_by(SigninRecord.signin_date.desc())
-        .limit(1)
-    ).scalar_one_or_none()
-    print(f"  最近一次签到: {last or '(从未签到)'}")
-
-
-    rows = db.execute(
-        select(CoinTransaction).where(
-            CoinTransaction.user_id == user.id,
-            CoinTransaction.biz_type == "signin",
-            CoinTransaction.ref_id == today.isoformat(),
-        )
-    ).scalars().all()
-    print(f"  coin_transaction(signin, 今天): {len(rows)} 条 / {sum(r.amount for r in rows)} 金币")
-
-    acc = db.get(CoinAccount, user.id)
-    if acc is None:
-        print("  coin_account: (无)")
-    else:
-        print(f"  coin_account: coin={acc.coin_balance} earned={acc.total_coin_earned}")
-
-    # 用 App 自己的 get_status 复核,而不是脚本里重算一遍规则 —— 这行就是客户端会看到的
-    st = crud_signin.get_status(db, user.id)
-    print(f"  [签到接口] can_claim={st.can_claim} today_signed={st.today_signed} "
-          f"今天第{st.today_cycle_day}档({st.today_coin}金币) 已连续{st.consecutive_days}天")
-
-
-def refund_today(db, user_id: int, today) -> None:
-    """退回今天签到(含膨胀)发的金币:删流水 + 扣余额。
-
-    不变量:coin_balance 必须恒等于流水总和。所以余额不够退时**整笔跳过**,而不是硬退成
-    负数、或夹到 0 —— 夹到 0 会吃掉用户在别处赚的金币,两种做法都会让余额和流水对不上。
-    """
-    rows = list(db.execute(
-        select(CoinTransaction).where(
-            CoinTransaction.user_id == user_id,
-            CoinTransaction.biz_type == "signin",
-            CoinTransaction.ref_id == today.isoformat(),
-        )
-    ).scalars().all())
-    if not rows:
-        return
-    acc = db.get(CoinAccount, user_id)
-    if acc is None:
-        return
-
-    # 从最近一笔往回退,退到余额兜不住为止:正常情况下今天只有一笔,整笔退掉 = 干净的撤销。
-    # 少数情况今天堆了多笔(上一轮测试时金币已被兑换、退不掉而留下的),这样也能保证
-    # 「本轮新发的那笔」一定被退掉 —— 否则每测一轮余额就永久多涨一档。
-    rows.sort(key=lambda r: r.id, reverse=True)
-    refundable: list[CoinTransaction] = []
-    total = 0
-    for r in rows:
-        if total + r.amount > acc.coin_balance:
-            break
-        refundable.append(r)
-        total += r.amount
-
-    for r in refundable:
-        db.delete(r)
-    if total:
-        acc.coin_balance -= total
-        acc.total_coin_earned = max(0, acc.total_coin_earned - total)
-        print(f"  已退回 {total} 金币({len(refundable)}/{len(rows)} 笔)")
-
-    stuck = len(rows) - len(refundable)
-    if stuck:
-        # 典型场景:签完就把金币兑换成现金了(exchange_out),这笔奖励已经变成 cash_balance_cents,
-        # 余额里已经没有它了。硬退会把余额退成负数 / 夹到 0 又会吃掉别处赚的金币,两者都会让账对不上。
-        print(f"  ⚠️ 还有 {stuck} 笔今天的签到流水退不掉(金币已被兑换/花掉,余额 {acc.coin_balance} 兜不住),"
-              f"原样保留 —— 硬退会让余额和流水总和对不上。")
-        print("     → 收益明细今天会多出几条签到记录,不影响签到功能测试;想彻底清干净用 reset_signin.py --with-coins。")
-
-
-def main() -> None:
-    parser = argparse.ArgumentParser(description="撤销用户今天的签到,让今天能重新签")
-    parser.add_argument("phone", nargs="?", default=DEFAULT_PHONE,
-                        help=f"手机号(默认 {DEFAULT_PHONE})")
-    parser.add_argument("--user-id", type=int, default=None, help="直接按 user_id 定位,优先于 phone")
-    parser.add_argument("--keep-coins", action="store_true",
-                        help="不退已发金币(余额会越测越高,且留下重复流水)")
-    parser.add_argument("--dry-run", action="store_true", help="预览,最后回滚不落库")
-    args = parser.parse_args()
-
-    if not args.dry_run and settings.APP_ENV != "dev":
-        raise SystemExit(f"APP_ENV={settings.APP_ENV},拒绝改库(只有 dev 能改;--dry-run 可任意环境)")
-
-    today = cn_today()
-    db = SessionLocal()
-    try:
-        user = resolve_user(db, args.phone, args.user_id)
-        print(f"DB: {settings.DATABASE_URL}")
-        print(f"用户: id={user.id} phone={user.phone}  今天(北京): {today}  keep_coins: {args.keep_coins}")
-        print_state(db, user, today, "before")
-
-        # 今天的签到记录 —— 只删今天,昨天及以前保留,重签后 streak 接着涨
-        rec = db.execute(
-            select(SigninRecord).where(
-                SigninRecord.user_id == user.id, SigninRecord.signin_date == today
-            )
-        ).scalar_one_or_none()
-        if rec is not None:
-            db.delete(rec)
-
-        if rec is None:
-            print("今天本来就没签到,无需处理。")
-            db.rollback()
-            return
-
-        # 退金币
-        if args.keep_coins:
-            print("(--keep-coins:保留已发金币,流水和余额不动)")
-        else:
-            refund_today(db, user.id, today)
-            # 注:更早流水的 balance_after 是当时的快照,不回改 —— 收益明细里历史行的
-            # 余额列会与现余额对不上,dev 测试库无妨。
-
-        # SessionLocal 是 autoflush=False,不 flush 的话下面 print_state 的 select
-        # 读到的还是删之前的旧行,"after" 会骗人
-        db.flush()
-        print_state(db, user, today, "after")
-
-        if args.dry_run:
-            db.rollback()
-            print("(dry-run:以上 after 为预览,已回滚,库没动)")
-            return
-        db.commit()
-        print(f"完成:{user.phone} 今天({today})可以重新签到了。"
-              f"提醒:App 内存状态不会自动同步,杀掉重进福利页(当天未签到)会重新自动弹签到弹窗。")
-    finally:
-        db.close()
-
-
-if __name__ == "__main__":
-    main()

scripts/seed_mock_llm_cost.py

@@ -1,248 +0,0 @@
-"""一次性 mock:造带 LLM token 成本的比价记录 + 配好 app_config 模型单价,用于测「管理后端」LLM 成本展示。
-
-覆盖 admin「比价记录」详情抽屉的「LLM 成本」展示分支:
-  • app_config.llm_token_price ← 写一条多模型单价(= 配置页「LLM 成本」卡片「已改」态,get_llm_prices 读它)
-  • comparison_record ← 造 5 条,逐条**复用生产的 compute_llm_cost + 与 _backfill_llm_calls 同款派生**
-    (llm_call_count/retry_count/input_tokens/output_tokens/llm_cost_yuan/llm_price_snapshot),
-    确保 mock 行 = 真实回填产出。5 条刻意覆盖:
-      ① 单模型真实样本(qwen3.5-flash ×4) → ¥0.006184(核对精确值)
-      ② 多模型(flash + plus)          → 快照含两个模型、各自 _source=per_model
-      ③ 未登记模型(deepseek-v3)        → 走 default,快照 _source=default
-      ④ 旧记录(有 token、无 cost)       → llm_cost_yuan=NULL → 前端回退「估算成本」
-      ⑤ 含 error 调用                    → error 那次跳过计费、retry_count+1
-
-记录挂到库里第一个真实用户(admin 列表能显示手机号);无用户则 user_id=NULL(孤儿行,admin 照样全看)。
-created_at 用北京 naive、最近几分钟内错开,详情列表倒序即 ①→⑤ 置顶。
-
-幂等:重跑先按 trace_id 前缀「MOCKLLM-」清旧再建。app_config 单价是 upsert(不随 --clean-only 删,
-因该 key 本就是本需求新增、无历史真实值;要改价直接去配置页或重跑本脚本)。
-
-  python -m scripts.seed_mock_llm_cost              # 造价格 + 5 条记录
-  python -m scripts.seed_mock_llm_cost --clean-only # 只清 MOCKLLM- 记录(保留单价)
-
-验收:admin「比价记录」→ 找 trace「MOCKLLM-」的 5 条 → 点开详情看「LLM 成本」:
-  ①②③⑤ 显示「实际·当时价」+ 价格快照;④ 显示「估算」。
-"""
-from __future__ import annotations
-
-import argparse
-import sys
-from datetime import datetime, timedelta, timezone
-
-from sqlalchemy import delete, select
-
-from app.db.session import SessionLocal
-from app.models.comparison import ComparisonRecord
-from app.models.user import User
-from app.repositories import app_config
-from app.services.llm_cost import compute_llm_cost
-
-if hasattr(sys.stdout, "reconfigure"):
-    sys.stdout.reconfigure(encoding="utf-8")  # Windows 控制台输出中文/¥
-
-_BJ = timezone(timedelta(hours=8))
-ID_PREFIX = "MOCKLLM-"
-
-# ── 写进 app_config 的模型单价(get_llm_prices 读它;配置页「LLM 成本」卡片可再改)──
-PRICE_CFG = {
-    "per_model": {
-        "qwen3.5-flash": {"input_per_1m": 0.8, "output_per_1m": 2.0},
-        "qwen3.5-plus": {"input_per_1m": 4.0, "output_per_1m": 12.0},
-    },
-    "default": {"input_per_1m": 3.0, "output_per_1m": 15.0},
-    "currency": "CNY",
-    "unit": "per_1m_tokens",
-}
-
-
-def _c(scene: str, model: str, pin: int, cout: int, error: str | None = None) -> dict:
-    """一条 llm_calls 明细,结构对齐真实 pricebot 归一后契约:
-    {scene, model, input_messages:[{role,content}], output, usage:{prompt/completion/total_tokens},
-     latency_ms, error}(详情抽屉会遍历 input_messages,缺了会崩)。error 的调用无 usage/output。"""
-    return {
-        "scene": scene,
-        "model": model,
-        "error": error,
-        "input_messages": [
-            {"role": "system", "content": f"你是比价助手,负责 {scene} 环节。"},
-            {"role": "user", "content": f"[mock] 请处理本次比价的 {scene} 任务。"},
-        ],
-        "output": None if error else f"[mock] {scene} 环节完成。",
-        "usage": None if error else {
-            "prompt_tokens": pin, "completion_tokens": cout, "total_tokens": pin + cout,
-        },
-        "latency_ms": 780,
-    }
-
-
-# ── 5 条记录蓝本:calls 决定成本;freeze=False 模拟旧记录(有 token 无 cost)──
-RECORDS = [
-    {
-        "label": "①单模型·真实样本",
-        "source": ("美团外卖", 4280), "best": ("京东秒送", 3680),
-        "store": "肯德基(建国路店)", "product": "疯狂星期四全家桶",
-        "info": "在京东秒送找到同款,到手价 ¥36.80,省 ¥6.00",
-        "freeze": True,
-        "calls": [
-            _c("store_match", "qwen3.5-flash", 1512, 22),
-            _c("dish_match", "qwen3.5-flash", 2111, 160),
-            _c("dish_match", "qwen3.5-flash", 1940, 142),
-            _c("summary", "qwen3.5-flash", 1325, 13),
-        ],
-    },
-    {
-        "label": "②多模型·flash+plus",
-        "source": ("淘宝闪购", 5900), "best": ("美团外卖", 5200),
-        "store": "瑞幸咖啡(国贸店)", "product": "生椰拿铁×2、丝绒拿铁",
-        "info": "在美团外卖找到同款,到手价 ¥52.00,省 ¥7.00",
-        "freeze": True,
-        "calls": [
-            _c("store_match", "qwen3.5-flash", 2000, 50),
-            _c("dish_match", "qwen3.5-flash", 1800, 40),
-            _c("reasoning", "qwen3.5-plus", 3000, 500),
-        ],
-    },
-    {
-        "label": "③未登记模型走 default",
-        "source": ("京东秒送", 3100), "best": ("美团外卖", 2650),
-        "store": "麦当劳(soho店)", "product": "麦辣鸡腿堡套餐",
-        "info": "在美团外卖找到同款,到手价 ¥26.50,省 ¥4.50",
-        "freeze": True,
-        "calls": [
-            _c("store_match", "deepseek-v3", 5000, 800),
-        ],
-    },
-    {
-        "label": "④旧记录·有token无成本(回退估算)",
-        "source": ("美团外卖", 3600), "best": ("淘宝闪购", 3200),
-        "store": "华莱士(双井店)", "product": "全鸡汉堡套餐",
-        "info": "在淘宝闪购找到同款,到手价 ¥32.00,省 ¥4.00",
-        "freeze": False,  # 模拟本需求上线前的老记录:llm_cost_yuan=NULL → 前端回退估算
-        "calls": [
-            _c("store_match", "qwen3.5-flash", 2000, 100),
-        ],
-    },
-    {
-        "label": "⑤含 error 调用(跳过计费)",
-        "source": ("淘宝闪购", 4100), "best": ("京东秒送", 3750),
-        "store": "海底捞(合生汇店)", "product": "番茄锅底、肥牛卷",
-        "info": "在京东秒送找到同款,到手价 ¥37.50,省 ¥3.50",
-        "freeze": True,
-        "calls": [
-            _c("store_match", "qwen3.5-flash", 0, 0, error="timeout"),
-            _c("store_match", "qwen3.5-flash", 1500, 30),
-        ],
-    },
-]
-
-_PLATFORM_ID = {  # 展示名 → 平台代号(comparison_results / source/best 列用)
-    "美团外卖": "meituan", "京东秒送": "jd", "淘宝闪购": "taobao",
-}
-
-
-def _naive_bj_now() -> datetime:
-    return datetime.now(_BJ).replace(tzinfo=None)
-
-
-def clean(db) -> int:
-    n = db.execute(
-        delete(ComparisonRecord).where(ComparisonRecord.trace_id.like(f"{ID_PREFIX}%"))
-    ).rowcount or 0
-    db.commit()
-    return n
-
-
-def _build_record(spec: dict, owner_id: int | None, created_at: datetime) -> tuple[ComparisonRecord, float | None]:
-    """按蓝本造一条记录,LLM 派生完全对齐 _backfill_llm_calls;返回 (记录, 冻结成本或 None)。"""
-    calls = spec["calls"]
-    src_name, src_cents = spec["source"]
-    best_name, best_cents = spec["best"]
-
-    # —— 与 _backfill_llm_calls 同款派生 ——
-    llm_call_count = len(calls)
-    retry_count = sum(1 for c in calls if c.get("error"))
-    input_tokens = sum((c.get("usage") or {}).get("prompt_tokens") or 0 for c in calls)
-    output_tokens = sum((c.get("usage") or {}).get("completion_tokens") or 0 for c in calls)
-    if spec["freeze"]:
-        cost, snapshot = compute_llm_cost(calls, PRICE_CFG)  # 复用生产纯函数
-    else:
-        cost, snapshot = None, None  # 旧记录:回填这段代码上线前就有,只有 token 没成本
-
-    rec = ComparisonRecord(
-        user_id=owner_id,
-        device_id=f"{ID_PREFIX.lower()}dev",
-        business_type="food",
-        trace_id=f"{ID_PREFIX}{spec['label'][0]}",  # ①..⑤ 各一,唯一
-        status="success",
-        source_platform_id=_PLATFORM_ID.get(src_name), source_platform_name=src_name,
-        source_price_cents=src_cents,
-        best_platform_id=_PLATFORM_ID.get(best_name), best_platform_name=best_name,
-        best_price_cents=best_cents,
-        saved_amount_cents=src_cents - best_cents,
-        is_source_best=False,
-        store_name=spec["store"],
-        product_names=spec["product"],
-        information=spec["info"],
-        items=[{"name": spec["product"], "qty": 1}],
-        comparison_results=[
-            {"platform_id": _PLATFORM_ID.get(src_name), "platform_name": src_name,
-             "price": src_cents / 100, "is_source": True, "rank": 2},
-            {"platform_id": _PLATFORM_ID.get(best_name), "platform_name": best_name,
-             "price": best_cents / 100, "is_source": False, "rank": 1},
-        ],
-        total_ms=90_000 + llm_call_count * 1000,
-        step_count=llm_call_count * 3,
-        llm_call_count=llm_call_count,
-        retry_count=retry_count,
-        input_tokens=input_tokens,
-        output_tokens=output_tokens,
-        llm_calls=calls,
-        llm_cost_yuan=cost,
-        llm_price_snapshot=snapshot,
-        created_at=created_at,
-    )
-    return rec, cost
-
-
-def seed(db) -> list[tuple[str, float | None]]:
-    app_config.set_value(db, "llm_token_price", PRICE_CFG, admin_id=None)  # upsert 单价
-    owner_id = db.execute(select(User.id).order_by(User.id).limit(1)).scalar()
-    base = _naive_bj_now()
-    out: list[tuple[str, float | None]] = []
-    for i, spec in enumerate(RECORDS):
-        rec, cost = _build_record(spec, owner_id, base - timedelta(minutes=i * 3))
-        db.add(rec)
-        out.append((spec["label"], cost))
-    db.commit()
-    return out, owner_id
-
-
-def main() -> None:
-    parser = argparse.ArgumentParser(description="造带 LLM 成本的比价记录 + app_config 模型单价(测管理后端)")
-    parser.add_argument("--clean-only", action="store_true", help="只清 MOCKLLM- 记录,不重建(保留单价)")
-    args = parser.parse_args()
-
-    db = SessionLocal()
-    try:
-        removed = clean(db)
-        if removed:
-            print(f"🧹 已清理旧 mock 记录 {removed} 条")
-        if args.clean_only:
-            print("✅ 仅清理,已完成(app_config 单价保留)。")
-            return
-
-        results, owner_id = seed(db)
-        print(f"\n✅ 已写入 app_config.llm_token_price(单价)+ {len(results)} 条比价记录"
-              f"(挂 user_id={owner_id or 'NULL(孤儿行)'})")
-        print("\n📋 每条冻结成本(admin 详情「LLM 成本」应显示):")
-        for label, cost in results:
-            shown = "NULL → 前端回退「估算」" if cost is None else f"¥{cost}"
-            print(f"  {label:<20} {shown}")
-        print("\n👉 验收:admin「比价记录」→ trace 搜「MOCKLLM-」→ 点开详情核对 LLM 成本 + 价格快照。")
-        print("   配置页「系统配置」→「福利页」Tab →「LLM 成本」卡片,单价应为「已改」态。")
-    finally:
-        db.close()
-
-
-if __name__ == "__main__":
-    main()

scripts/seed_mock_withdraws.py

@@ -7,7 +7,7 @@
   - 配套双分录现金流水(withdraw / withdraw_refund / exchange_in)+ 账户余额,
     让顶部「账本校验」保持绿色、详情抽屉的现金余额/流水也真实。
 
-约束:withdraw_order 有部分唯一索引(同一 user 在 reviewing/pending 最多 1 单),
+约束:同一 user 可有多笔 reviewing/pending 单,每笔 out_bill_no 唯一且独立对账,
 本脚本每个 mock 用户至多 1 个活动单,满足约束。
 
 幂等:每次运行先按固定 mock 手机号清掉旧 mock 再重建。仅清理用 --clean-only。

tests/test_ad_reward.py

@@ -356,272 +356,3 @@ def test_callback_disabled_returns_503(client, monkeypatch) -> None:
     # 503 发生在验签/发奖之前,不需要真实用户
     r = _callback(client, _signed(1, "trans_disabled"))
     assert r.status_code == 503, r.text
-
-
-# ===== 按 ad_session_id 查权威发奖结果(GET /reward-result/{ad_session_id})=====
-# 客户端看完广告轮询它拿弹窗金额:只认 status='granted' 且 coin>0,其余一律不弹。
-
-
-def _reward_result(client, token: str, session_id: str):
-    return client.get(f"/api/v1/ad/reward-result/{session_id}", headers=_auth(token))
-
-
-def _session_extra(session_id: str, **kv: str) -> str:
-    return json.dumps({"ad_session_id": session_id, **kv})
-
-
-def test_reward_result_pending_when_s2s_not_arrived(client) -> None:
-    """S2S 还没回调 → 200 + pending(**不是 404**),客户端据此继续轮询。"""
-    token = _login(client, "13800003601")
-
-    r = _reward_result(client, token, "sess-not-yet-arrived")
-    assert r.status_code == 200, r.text
-    assert r.json() == {
-        "ad_session_id": "sess-not-yet-arrived",
-        "status": "pending",
-        "coin": None,
-        # 没记录 → 连属于哪一轮都不知道,累计值一并为 null(不是 0,0 会被读成"本轮没赚到")
-        "round_coin": None,
-    }
-
-
-def test_reward_result_returns_granted_coin(client) -> None:
-    """S2S 发奖后按会话查 → granted + 本次真实到账额(与钱包入账一致)。"""
-    phone = "13800003602"
-    token = _login(client, phone)
-    uid = _user_id(phone)
-    session_id = "sess-granted-1"
-
-    r = _callback(
-        client,
-        _signed(uid, "trans_rr_1", ecpm="200", extra=_session_extra(session_id)),
-    )
-    assert r.json() == {"is_verify": True, "reason": 0}
-
-    expected = calculate_ad_reward_coin("200", 1)
-    body = _reward_result(client, token, session_id).json()
-    assert body["status"] == "granted"
-    assert body["coin"] == expected
-    # 弹窗金额必须等于真实入账,这正是本接口存在的意义(不用余额差估算)
-    assert _coin_balance(client, token) == expected
-
-
-def test_reward_result_prefers_granted_over_earlier_noshow(client) -> None:
-    """竞态:客户端先报 closed_early、S2S 随后才到 → 同一会话两条记录,必须返回 granted 那条。
-
-    只按 created_at 取最近一条是不够的(SQLite 下两条可能同一时间戳),故仓储层显式优先 granted。
-    """
-    phone = "13800003603"
-    token = _login(client, phone)
-    uid = _user_id(phone)
-    session_id = "sess-race-noshow"
-
-    # 1) 客户端以为没发奖,先留痕
-    r = client.post(
-        "/api/v1/ad/reward-noshow",
-        json={"ad_session_id": session_id, "watched_seconds": 3},
-        headers=_auth(token),
-    )
-    assert r.status_code == 200, r.text
-    assert r.json()["status"] == "closed_early"
-    assert _reward_result(client, token, session_id).json()["status"] == "closed_early"
-
-    # 2) S2S 姗姗来迟,真发了钱
-    _callback(client, _signed(uid, "trans_rr_race", ecpm="200", extra=_session_extra(session_id)))
-
-    body = _reward_result(client, token, session_id).json()
-    assert body["status"] == "granted"
-    assert body["coin"] == calculate_ad_reward_coin("200", 1)
-
-
-def test_reward_result_capped_reports_zero_not_popup(client) -> None:
-    """达每日上限 → capped + coin=0;客户端不弹「获得 0 金币」。"""
-    phone = "13800003604"
-    token = _login(client, phone)
-    uid = _user_id(phone)
-    session_id = "sess-capped-1"
-
-    db = SessionLocal()
-    try:
-        db.add(
-            AdRewardRecord(
-                trans_id="trans_rr_capped", user_id=_user_id(phone), coin=0, status="capped",
-                reward_scene="reward_video", ad_session_id=session_id, reward_date="2026-07-17",
-            )
-        )
-        db.commit()
-    finally:
-        db.close()
-    assert uid  # 记录挂在该用户名下
-
-    body = _reward_result(client, token, session_id).json()
-    assert body["status"] == "capped"
-    assert body["coin"] == 0
-
-
-def test_reward_result_scoped_to_owner(client) -> None:
-    """别人的会话查不到(按 user_id 收窄)→ pending,不泄漏他人发奖结果。"""
-    phone_a = "13800003605"
-    token_a = _login(client, phone_a)
-    uid_a = _user_id(phone_a)
-    token_b = _login(client, "13800003606")
-    session_id = "sess-owner-only"
-
-    _callback(client, _signed(uid_a, "trans_rr_owner", ecpm="200", extra=_session_extra(session_id)))
-
-    assert _reward_result(client, token_a, session_id).json()["status"] == "granted"
-    assert _reward_result(client, token_b, session_id).json()["status"] == "pending"
-
-
-def test_reward_result_requires_auth(client) -> None:
-    """无 Bearer → 401,不裸奔。"""
-    assert client.get("/api/v1/ad/reward-result/sess-anon-1").status_code == 401
-
-
-# ===== 膨胀轮累计(boost_round_id → reward-result.round_coin)=====
-# 不变量:弹窗数字 == 本轮实际到账之和 == 余额涨幅。三者对不上用户就认为少发了钱。
-
-
-def _round_extra(session_id: str, round_id: str | None = None, **kv: str) -> str:
-    data = {"ad_session_id": session_id, **kv}
-    if round_id is not None:
-        data["boost_round_id"] = round_id
-    return json.dumps(data)
-
-
-def test_round_coin_accumulates_across_ads_in_same_round(client) -> None:
-    """一轮连看两条 → round_coin 逐条累计,且等于余额涨幅(第七节验收 1、2 步)。"""
-    phone = "13800003701"
-    token = _login(client, phone)
-    uid = _user_id(phone)
-    round_id = "b7e1c93a4f6d802b"
-
-    _callback(client, _signed(
-        uid, "trans_round_1", ecpm="200", extra=_round_extra("sess-r1-a", round_id)))
-    first = calculate_ad_reward_coin("200", 1)
-    body = _reward_result(client, token, "sess-r1-a").json()
-    assert body["coin"] == first
-    assert body["round_coin"] == first  # 第 1 条:本轮累计 == 本条
-
-    _callback(client, _signed(
-        uid, "trans_round_2", ecpm="200", extra=_round_extra("sess-r1-b", round_id)))
-    second = calculate_ad_reward_coin("200", 2)  # LT 因子递减,第 2 条比第 1 条少
-    body = _reward_result(client, token, "sess-r1-b").json()
-    assert body["coin"] == second
-    assert body["round_coin"] == first + second  # 累计 = 两条之和
-
-    # 弹窗数字必须等于真实余额涨幅 —— 这条不变量是整个方案的目的
-    assert _coin_balance(client, token) == first + second
-
-
-def test_new_round_restarts_accumulation(client) -> None:
-    """换新轮 id → round_coin 从头累计,不接着上一轮往上加(第七节最后一句验收)。"""
-    phone = "13800003702"
-    token = _login(client, phone)
-    uid = _user_id(phone)
-
-    _callback(client, _signed(uid, "trans_r2_old", ecpm="200", extra=_round_extra("sess-r2-a", "round-old")))
-    old = _reward_result(client, token, "sess-r2-a").json()["round_coin"]
-    assert old > 0
-
-    _callback(client, _signed(uid, "trans_r2_new", ecpm="200", extra=_round_extra("sess-r2-b", "round-new")))
-    body = _reward_result(client, token, "sess-r2-b").json()
-    assert body["round_coin"] == body["coin"]  # 新轮 = 只有本条
-    assert body["round_coin"] != old + body["coin"]
-
-
-def test_round_coin_null_without_round_id(client) -> None:
-    """extra 没带 boost_round_id(老客户端 / GroMore 丢字段)→ round_coin=null,客户端退回显示单条。"""
-    phone = "13800003703"
-    token = _login(client, phone)
-    uid = _user_id(phone)
-
-    _callback(client, _signed(uid, "trans_r3", ecpm="200", extra=_round_extra("sess-r3-noround", None)))
-    body = _reward_result(client, token, "sess-r3-noround").json()
-    assert body["coin"] == calculate_ad_reward_coin("200", 1)
-    assert body["round_coin"] is None
-
-
-def test_round_coin_null_when_pending(client) -> None:
-    """S2S 未到账 → 没有记录 → 连轮 id 都不知道,round_coin 也是 null(不是 0)。"""
-    token = _login(client, "13800003704")
-    body = _reward_result(client, token, "sess-r4-pending").json()
-    assert body == {
-        "ad_session_id": "sess-r4-pending",
-        "status": "pending",
-        "coin": None,
-        "round_coin": None,
-    }
-
-
-def test_round_coin_returned_on_capped(client) -> None:
-    """撞每日上限那条不是 granted,但 round_coin **仍返本轮累计**(该条按 0 计)。
-
-    客户端的限额 toast 要显示前面几条已到账的总额,不能是空。
-    """
-    phone = "13800003705"
-    token = _login(client, phone)
-    uid = _user_id(phone)
-    round_id = "round-capped"
-
-    _callback(client, _signed(uid, "trans_cap_ok", ecpm="200", extra=_round_extra("sess-cap-a", round_id)))
-    earned = _reward_result(client, token, "sess-cap-a").json()["round_coin"]
-    assert earned > 0
-
-    # 手插一条同轮的 capped 记录(跑满 500 次太慢),模拟第 N 条撞上限
-    db = SessionLocal()
-    try:
-        db.add(AdRewardRecord(
-            trans_id="trans_cap_hit", user_id=uid, coin=0, status="capped",
-            reward_scene="reward_video", ad_session_id="sess-cap-b",
-            reward_date="2026-07-20", boost_round_id=round_id,
-        ))
-        db.commit()
-    finally:
-        db.close()
-
-    body = _reward_result(client, token, "sess-cap-b").json()
-    assert body["status"] == "capped"
-    assert body["coin"] == 0          # 这条没发钱
-    assert body["round_coin"] == earned  # 但本轮累计照常返回
-
-
-def test_round_coin_scoped_to_owner(client) -> None:
-    """轮 id 是客户端生成的,不能跨用户信任:拿别人的轮 id 查不到别人的金币。"""
-    phone_a = "13800003706"
-    token_a = _login(client, phone_a)
-    uid_a = _user_id(phone_a)
-    phone_b = "13800003707"
-    token_b = _login(client, phone_b)
-    uid_b = _user_id(phone_b)
-    shared_round = "round-collision"
-
-    _callback(client, _signed(uid_a, "trans_own_a", ecpm="200", extra=_round_extra("sess-own-a", shared_round)))
-    a_total = _reward_result(client, token_a, "sess-own-a").json()["round_coin"]
-
-    # B 用同一个轮 id(伪造或碰撞)看一条:B 的累计里不能混进 A 的钱
-    _callback(client, _signed(uid_b, "trans_own_b", ecpm="200", extra=_round_extra("sess-own-b", shared_round)))
-    b_body = _reward_result(client, token_b, "sess-own-b").json()
-    assert b_body["round_coin"] == b_body["coin"]
-    assert b_body["round_coin"] < a_total + b_body["coin"]
-
-
-def test_test_grant_accepts_boost_round_id(client, monkeypatch) -> None:
-    """debug 的 test-grant 不经 S2S、拿不到 mediaExtra,轮 id 由 body 补 → 本地也能验累计。"""
-    monkeypatch.setattr(settings, "AD_REWARD_TEST_GRANT_ENABLED", True)
-    token = _login(client, "13800003708")
-    round_id = "round-testgrant"
-
-    coins = []
-    for i in range(2):
-        r = client.post(
-            "/api/v1/ad/test-grant",
-            json={"reward_scene": "reward_video", "boost_round_id": round_id,
-                  "ad_session_id": f"sess-tg-{i}-padding"},
-            headers=_auth(token),
-        )
-        assert r.status_code == 200, r.text
-        coins.append(r.json()["coin"])
-
-    body = _reward_result(client, token, "sess-tg-1-padding").json()
-    assert body["round_coin"] == sum(coins)

tests/test_auth.py

@@ -107,67 +107,6 @@ def test_sms_send_device_ip_rate_limit(client, monkeypatch) -> None:
     assert r.status_code == 200, r.text
 
 
-def test_sms_send_cooldown_reject_not_counted(client, monkeypatch) -> None:
-    """发码额度只算「成功发码」:被单号 60s 冷却挡下的重发(429)不占设备额度。
-    做法:同号狂发只成功 1 次、其余被冷却挡下;把小时额度设 2,证明换号后仍能再成功发 1 次
-    —— 若冷却重发也计数,额度早被那几次耗尽。"""
-    from app.api.v1 import auth
-    from app.core import ratelimit
-
-    monkeypatch.setattr(ratelimit.settings, "RATE_LIMIT_ENABLED", True)
-    monkeypatch.setattr(auth, "SMS_SEND_MAX_PER_HOUR_PER_DEVICE", 2)
-    ratelimit._buckets.clear()
-
-    device = "dev-cooldown"
-    phone_a = "13710137000"
-    # 首发成功(小时闸计 1/2)
-    assert client.post(
-        "/api/v1/auth/sms/send", json={"phone": phone_a, "device_id": device}
-    ).status_code == 200
-    # 同号连发 3 次:都被单号 60s 冷却挡下 → 429,且**不占**设备额度
-    for _ in range(3):
-        r = client.post(
-            "/api/v1/auth/sms/send", json={"phone": phone_a, "device_id": device}
-        )
-        assert r.status_code == 429, r.text
-    # 换号再发:设备额度只用了 1/2(冷却那几次没算)→ 仍放行(计到 2/2)
-    assert client.post(
-        "/api/v1/auth/sms/send", json={"phone": "13710137001", "device_id": device}
-    ).status_code == 200
-    # 又换号:此时小时闸已 2/2 → 429(反证成功发码确实各计了 1)
-    r = client.post(
-        "/api/v1/auth/sms/send", json={"phone": "13710137002", "device_id": device}
-    )
-    assert r.status_code == 429, r.text
-
-
-def test_sms_send_daily_cap(client, monkeypatch) -> None:
-    """每天发码上限(设备 + IP):成功发码累计到日上限即 429(用不同手机号绕开单号冷却)。
-    抬高小时闸单独测日闸;超限文案含「今日」以便前端提示明天再来。"""
-    from app.api.v1 import auth
-    from app.core import ratelimit
-
-    monkeypatch.setattr(ratelimit.settings, "RATE_LIMIT_ENABLED", True)
-    monkeypatch.setattr(auth, "SMS_SEND_MAX_PER_HOUR_PER_DEVICE", 100)  # 抬高小时闸,不干扰
-    monkeypatch.setattr(auth, "SMS_SEND_MAX_PER_DAY_PER_DEVICE", 3)
-    ratelimit._buckets.clear()
-
-    device = "dev-daily"
-    for i in range(3):
-        r = client.post(
-            "/api/v1/auth/sms/send",
-            json={"phone": f"13720137{i:03d}", "device_id": device},
-        )
-        assert r.status_code == 200, f"第 {i + 1} 次应放行: {r.text}"
-    # 第 4 次:同设备同 IP 当日超限 → 429
-    r = client.post(
-        "/api/v1/auth/sms/send",
-        json={"phone": "13720137999", "device_id": device},
-    )
-    assert r.status_code == 429, r.text
-    assert "今日" in r.json()["detail"]
-
-
 def test_sms_login_device_ip_rate_limit(client, monkeypatch) -> None:
     """防刷:同一设备(device_id) + 同一 IP 每小时最多 SMS_LOGIN_MAX_PER_HOUR 次登录尝试,超出 429。
     conftest 默认 RATE_LIMIT_ENABLED=false(内存计数跨用例累加),本用例临时打开并清空计数隔离。"""

tests/test_invite_cash_withdraw.py

@@ -128,8 +128,7 @@ def test_invite_cash_reject_refunds_invite_account(client, monkeypatch) -> None:
 
 
 def test_two_accounts_withdraw_independent(client, monkeypatch) -> None:
-    """两账户各提各的不串:先提 invite_cash(拒绝结清),再提 cash,各扣各账户。
-    注:一个用户同一时间只能一个活跃提现单(跨账户),故第二笔需先结清第一笔。"""
+    """两账户各提各的不串:invite_cash 与 coin_cash 可同时保留 reviewing 单。"""
     _patch_userinfo(monkeypatch, "openid_ic_3")
     token = _login(client, "13800004003")
     _seed_balances(client, token, "13800004003", cash=400, invite_cash=500)
@@ -140,17 +139,17 @@ def test_two_accounts_withdraw_independent(client, monkeypatch) -> None:
         json={"amount_cents": 200, "source": "invite_cash"},
         headers=_auth(token),
     )
-    _reject(r1.json()["out_bill_no"])    # 退回 invite_cash + 结清活跃单
+    assert r1.status_code == 200 and r1.json()["status"] == "reviewing", r1.text
     r2 = client.post(
         "/api/v1/wallet/withdraw",
         # 50 分 = 0.5 元档(7-9 起 coin_cash 只能提预设档位)
         json={"amount_cents": 50, "source": "coin_cash"},
         headers=_auth(token),
     )
-    assert r2.json()["status"] == "reviewing"
+    assert r2.status_code == 200 and r2.json()["status"] == "reviewing", r2.text
 
     cash, invite_cash = _balances(client, token)
-    assert invite_cash == 500    # 已退回
+    assert invite_cash == 300    # invite_cash 单仍待审核,已扣 200
     assert cash == 350           # 扣了 cash 50
 
 

tests/test_llm_cost.py

@@ -1,184 +0,0 @@
-"""LLM 调用成本计算 compute_llm_cost:按模型分桶累加 token × 单价;error/无 usage 跳过。"""
-from __future__ import annotations
-
-from app.services.llm_cost import compute_llm_cost
-
-_PRICE = {
-    "per_model": {"qwen3.5-flash": {"input_per_1m": 0.8, "output_per_1m": 2.0}},
-    "default": {"input_per_1m": 3.0, "output_per_1m": 15.0},
-}
-
-
-def test_sums_per_model_single_model():
-    # 真实样本:4 次 qwen3.5-flash;Σprompt=6888、Σcompletion=337
-    calls = [
-        {"model": "qwen3.5-flash", "error": None, "usage": {"prompt_tokens": 1512, "completion_tokens": 22}},
-        {"model": "qwen3.5-flash", "error": None, "usage": {"prompt_tokens": 2111, "completion_tokens": 160}},
-        {"model": "qwen3.5-flash", "error": None, "usage": {"prompt_tokens": 1940, "completion_tokens": 142}},
-        {"model": "qwen3.5-flash", "error": None, "usage": {"prompt_tokens": 1325, "completion_tokens": 13}},
-    ]
-    cost, snapshot = compute_llm_cost(calls, _PRICE)
-    # 6888/1e6*0.8 + 337/1e6*2.0 = 0.0055104 + 0.000674 = 0.0061844 → round(6)
-    assert cost == 0.006184
-    assert snapshot == {
-        "mode": "per_model",
-        "prices": {
-            "qwen3.5-flash": {"input_per_1m": 0.8, "output_per_1m": 2.0, "_source": "per_model"},
-        },
-    }
-
-
-def test_multi_model_prices_each_bucket_separately():
-    calls = [
-        {"model": "qwen3.5-flash", "error": None, "usage": {"prompt_tokens": 1_000_000, "completion_tokens": 0}},
-        {"model": "gpt-x", "error": None, "usage": {"prompt_tokens": 0, "completion_tokens": 1_000_000}},
-    ]
-    price = {
-        "per_model": {
-            "qwen3.5-flash": {"input_per_1m": 0.8, "output_per_1m": 2.0},
-            "gpt-x": {"input_per_1m": 10.0, "output_per_1m": 30.0},
-        },
-        "default": {"input_per_1m": 3.0, "output_per_1m": 15.0},
-    }
-    cost, snap = compute_llm_cost(calls, price)
-    assert cost == 30.8  # qwen 1M入×0.8=0.8 + gpt-x 1M出×30=30.0
-    assert set(snap["prices"]) == {"qwen3.5-flash", "gpt-x"}
-
-
-def test_unknown_model_falls_back_to_default():
-    calls = [{"model": "mystery", "error": None, "usage": {"prompt_tokens": 1_000_000, "completion_tokens": 0}}]
-    price = {"per_model": {}, "default": {"input_per_1m": 3.0, "output_per_1m": 15.0}}
-    cost, snap = compute_llm_cost(calls, price)
-    assert cost == 3.0
-    assert snap["prices"]["mystery"]["_source"] == "default"
-
-
-def test_unpriced_model_marked_and_zero_cost():
-    calls = [{"model": "mystery", "error": None, "usage": {"prompt_tokens": 1_000_000, "completion_tokens": 999}}]
-    cost, snap = compute_llm_cost(calls, {"per_model": {}})  # 无 default
-    assert cost == 0.0
-    assert snap["prices"]["mystery"]["unpriced"] is True
-
-
-def test_error_and_missing_usage_calls_skipped():
-    calls = [
-        {"model": "qwen3.5-flash", "error": "boom", "usage": {"prompt_tokens": 9_999_999, "completion_tokens": 9_999_999}},
-        {"model": "qwen3.5-flash", "error": None, "usage": None},  # 无 usage
-        {"model": "qwen3.5-flash", "error": None, "usage": {"prompt_tokens": 1_000_000, "completion_tokens": 0}},
-    ]
-    cost, _ = compute_llm_cost(calls, _PRICE)
-    assert cost == 0.8  # 只有第 3 条计入
-
-
-def test_empty_or_all_error_returns_none():
-    assert compute_llm_cost([], _PRICE) == (None, None)
-    assert compute_llm_cost(None, _PRICE) == (None, None)
-    all_error = [{"model": "x", "error": "boom", "usage": {"prompt_tokens": 100, "completion_tokens": 100}}]
-    assert compute_llm_cost(all_error, _PRICE) == (None, None)
-
-
-def test_malformed_price_entry_is_treated_as_unpriced_not_raised():
-    # 手改配置页可能存出残缺/非法单价(缺 output_per_1m、非 dict);不能抛异常连累 token 回填。
-    calls = [
-        {"model": "bad-a", "error": None, "usage": {"prompt_tokens": 1_000_000, "completion_tokens": 5}},
-        {"model": "bad-b", "error": None, "usage": {"prompt_tokens": 1_000_000, "completion_tokens": 5}},
-        {"model": "ok", "error": None, "usage": {"prompt_tokens": 1_000_000, "completion_tokens": 0}},
-    ]
-    price = {
-        "per_model": {
-            "bad-a": {"input_per_1m": 0.8},   # 缺 output_per_1m
-            "bad-b": 5,                        # 非 dict
-            "ok": {"input_per_1m": 3.0, "output_per_1m": 15.0},
-        },
-    }
-    cost, snap = compute_llm_cost(calls, price)  # 不得抛异常
-    assert cost == 3.0  # 只有 ok(1M 入 × 3.0)计入;两个残缺项按 unpriced
-    assert snap["prices"]["bad-a"].get("unpriced") is True
-    assert snap["prices"]["bad-b"].get("unpriced") is True
-
-
-def test_get_llm_prices_falls_back_to_default_then_uses_override():
-    from app.db.session import SessionLocal
-    from app.models.app_config import AppConfig
-    from app.repositories import app_config
-    from app.services.llm_cost import get_llm_prices
-
-    db = SessionLocal()
-    try:
-        # 无 override → CONFIG_DEFS 默认(含 per_model / default)
-        prices = get_llm_prices(db)
-        assert "per_model" in prices and "default" in prices
-        # 有 override → 用 DB 值
-        app_config.set_value(
-            db, "llm_token_price",
-            {"per_model": {"m": {"input_per_1m": 1.0, "output_per_1m": 2.0}},
-             "default": {"input_per_1m": 0.0, "output_per_1m": 0.0}},
-            admin_id=1,
-        )
-        assert get_llm_prices(db)["per_model"]["m"]["input_per_1m"] == 1.0
-    finally:
-        row = db.get(AppConfig, "llm_token_price")
-        if row is not None:
-            db.delete(row)
-            db.commit()
-        db.close()
-
-
-def test_backfill_llm_calls_stores_cost_and_snapshot(monkeypatch):
-    from datetime import UTC, datetime
-
-    from app.api.v1 import compare_record
-    from app.db.session import SessionLocal
-    from app.models.app_config import AppConfig
-    from app.models.comparison import ComparisonRecord
-    from app.repositories import app_config
-
-    sample = [
-        {"model": "qwen3.5-flash", "error": None, "usage": {"prompt_tokens": 1512, "completion_tokens": 22}},
-        {"model": "qwen3.5-flash", "error": None, "usage": {"prompt_tokens": 2111, "completion_tokens": 160}},
-        {"model": "qwen3.5-flash", "error": None, "usage": {"prompt_tokens": 1940, "completion_tokens": 142}},
-        {"model": "qwen3.5-flash", "error": None, "usage": {"prompt_tokens": 1325, "completion_tokens": 13}},
-    ]
-    monkeypatch.setattr(compare_record, "fetch_llm_calls", lambda trace_id: sample)
-
-    db = SessionLocal()
-    try:
-        app_config.set_value(
-            db, "llm_token_price",
-            {"per_model": {"qwen3.5-flash": {"input_per_1m": 0.8, "output_per_1m": 2.0}},
-             "default": {"input_per_1m": 3.0, "output_per_1m": 15.0}},
-            admin_id=1,
-        )
-        rec = ComparisonRecord(
-            trace_id="llmcost-bf-1", status="success",
-            created_at=datetime.now(UTC).replace(tzinfo=None),
-        )
-        db.add(rec)
-        db.commit()
-        rid = rec.id
-    finally:
-        db.close()
-
-    compare_record._backfill_llm_calls(rid, "llmcost-bf-1")  # 独立 session 内回填
-
-    db = SessionLocal()
-    try:
-        rec = db.get(ComparisonRecord, rid)
-        assert rec.llm_cost_yuan == 0.006184
-        assert rec.llm_price_snapshot["prices"]["qwen3.5-flash"]["input_per_1m"] == 0.8
-        assert rec.input_tokens == 6888  # 现有 token 派生仍在
-    finally:
-        db.delete(db.get(ComparisonRecord, rid))
-        row = db.get(AppConfig, "llm_token_price")
-        if row is not None:
-            db.delete(row)
-        db.commit()
-        db.close()
-
-
-def test_admin_detail_schema_exposes_llm_cost_fields():
-    from app.admin.schemas.comparison import AdminComparisonDetail
-
-    fields = AdminComparisonDetail.model_fields
-    assert "llm_cost_yuan" in fields
-    assert "llm_price_snapshot" in fields

tests/test_ratelimit.py

@@ -1,57 +0,0 @@
-"""ratelimit 内存桶过期清理(GC)测试。
-
-回归重点:_buckets 是**全局共享**、混着不同窗口(60s 广告 / 3600s 登录 / 86400s 日闸)的 key。
-GC 必须按【每个 key 自己存的 window_sec】判过期,而不是当前调用方的窗口 —— 否则高频的 60s 端点
-触发 GC 时会把本该存活更久的 3600s/86400s 计数(如短信日闸)一并删掉,使其被反复清零、限流失效。
-用 monkeypatch 把 _GC_THRESHOLD 调 0 强制每次都扫,免造上万条(仿 test_auth 里对 sms._GC_THRESHOLD 的做法)。
-"""
-from __future__ import annotations
-
-from app.core import ratelimit
-
-
-def test_purge_expired_respects_each_key_own_window(monkeypatch) -> None:
-    """短窗口(60s)触发的 GC 只删真正过期的 key,不得删掉仍在自身窗口内的长窗口 key。"""
-    monkeypatch.setattr(ratelimit, "_GC_THRESHOLD", 0)  # 强制每次都扫
-    ratelimit._buckets.clear()
-
-    now = 1_000_000.0
-    # 日闸:100s 前开窗、window=86400 → 远未过期,必须保留
-    ratelimit._buckets["sms-send-device-daily:D:IP"] = (now - 100, 7, 86400.0)
-    # 登录:1800s、window=3600 → 未过期,保留
-    ratelimit._buckets["sms-login-device:D:IP"] = (now - 1800, 2, 3600.0)
-    # 广告:120s、window=60 → 已过期,应删
-    ratelimit._buckets["ad-watch-report:IP2"] = (now - 120, 3, 60.0)
-
-    ratelimit._purge_expired(now)
-
-    assert "sms-send-device-daily:D:IP" in ratelimit._buckets
-    assert "sms-login-device:D:IP" in ratelimit._buckets
-    assert "ad-watch-report:IP2" not in ratelimit._buckets
-
-
-def test_purge_expired_keeps_long_window_key_older_than_short_window(monkeypatch) -> None:
-    """反证旧 bug:日闸 key 已老于 3600s,旧代码在 60s/3600s 端点触发 GC 时会误删它;
-    现在按自身 86400s 窗口判 → 未过期 → 必须保留。"""
-    monkeypatch.setattr(ratelimit, "_GC_THRESHOLD", 0)
-    ratelimit._buckets.clear()
-
-    now = 2_000_000.0
-    # 3700s 前开窗(> 1 小时),但 window=86400 → 未过期
-    ratelimit._buckets["sms-send-device-daily:D:IP"] = (now - 3700, 20, 86400.0)
-
-    ratelimit._purge_expired(now)
-
-    assert "sms-send-device-daily:D:IP" in ratelimit._buckets
-
-
-def test_purge_expired_noop_below_threshold(monkeypatch) -> None:
-    """未超阈值时不扫(即便有过期 key 也不动),避免每次请求都 O(n) 扫全表。"""
-    monkeypatch.setattr(ratelimit, "_GC_THRESHOLD", 10)
-    ratelimit._buckets.clear()
-
-    now = 3_000_000.0
-    ratelimit._buckets["stale:IP"] = (now - 999, 1, 60.0)  # 早过期,但没超阈值
-    ratelimit._purge_expired(now)
-
-    assert "stale:IP" in ratelimit._buckets  # 桶数没超阈值 → 不清理

tests/test_welfare.py

@@ -11,6 +11,7 @@ from app.core.rewards import (
     COIN_PER_CENT,
     COIN_PER_YUAN,
     MIN_EXCHANGE_COIN,
+    SIGNIN_BOOST_COIN,
     SIGNIN_REWARDS,
     TASK_ENABLE_NOTIFICATION,
     TASK_REWARDS,
@@ -105,6 +106,48 @@ def test_signin_flow(client) -> None:
     assert txn["balance_after"] == SIGNIN_REWARDS[0]
 
 
+def test_signin_boost_flow(client) -> None:
+    """签到后看广告膨胀 → S2S 固定补发 2000 金币,每天只能膨胀一次。"""
+    phone = "13800001011"
+    token = _login(client, phone)
+
+    r = client.post("/api/v1/signin/boost", json={}, headers=_auth(token))
+    assert r.status_code == 409
+
+    r = client.post("/api/v1/signin", headers=_auth(token))
+    assert r.status_code == 200, r.text
+    first_coin = r.json()["coin_awarded"]
+
+    with SessionLocal() as db:
+        user = get_user_by_phone(db, phone)
+        assert user is not None
+        uid = user.id
+
+    extra = json.dumps({"reward_scene": "signin_boost", "ad_session_id": "signin-session-1"})
+    r = client.get(
+        "/api/v1/ad/pangle-callback",
+        params=_signed_ad_callback(uid, "signin-boost-trans-1", extra=extra, ecpm="200"),
+    )
+    assert r.status_code == 200, r.text
+    assert r.json() == {"is_verify": True, "reason": 0}
+
+    r = client.post("/api/v1/signin/boost", json={"ad_ref_id": "signin-boost-trans-1"}, headers=_auth(token))
+    assert r.status_code == 200, r.text
+    body = r.json()
+    assert body["coin_awarded"] == SIGNIN_BOOST_COIN
+    assert body["coin_balance"] == first_coin + SIGNIN_BOOST_COIN
+
+    r = client.get(
+        "/api/v1/ad/pangle-callback",
+        params=_signed_ad_callback(uid, "signin-boost-trans-2", extra=extra, ecpm="200"),
+    )
+    assert r.status_code == 200, r.text
+
+    r = client.get("/api/v1/wallet/coin-transactions", headers=_auth(token))
+    types = [t["biz_type"] for t in r.json()["items"]]
+    assert "signin" in types
+    assert "signin_boost" in types
+
 
 def test_task_claim_flow(client) -> None:
     """打开消息提醒=可重复任务:每次领取金额减半(750/375/188),claimed 恒 False,余额累加。"""

tests/test_withdraw.py

@@ -268,6 +268,34 @@ def test_withdraw_idempotent_same_bill_no(client, monkeypatch) -> None:
     assert sum(1 for t in r.json()["items"] if t["biz_type"] == "withdraw") == 1
 
 
+def test_withdraw_allows_multiple_active_orders_with_distinct_bill_no(client, monkeypatch) -> None:
+    """不同 out_bill_no 是不同提现意图,允许同时处于 reviewing；每笔分别扣款建单。"""
+    _patch_userinfo(monkeypatch, "openid_multi_active")
+    token = _login(client, "13800002017")
+    _seed_cash(client, token, "13800002017", 200)
+    client.post("/api/v1/wallet/bind-wechat", json={"code": "c"}, headers=_auth(token))
+
+    r1 = client.post(
+        "/api/v1/wallet/withdraw",
+        json={"amount_cents": 50, "out_bill_no": "multi_active_bill_1"},
+        headers=_auth(token),
+    )
+    r2 = client.post(
+        "/api/v1/wallet/withdraw",
+        json={"amount_cents": 50, "out_bill_no": "multi_active_bill_2"},
+        headers=_auth(token),
+    )
+
+    assert r1.status_code == 200 and r1.json()["status"] == "reviewing", r1.text
+    assert r2.status_code == 200 and r2.json()["status"] == "reviewing", r2.text
+    assert r1.json()["out_bill_no"] != r2.json()["out_bill_no"]
+
+    account = client.get("/api/v1/wallet/account", headers=_auth(token)).json()
+    assert account["cash_balance_cents"] == 100
+    txns = client.get("/api/v1/wallet/cash-transactions", headers=_auth(token)).json()["items"]
+    assert sum(1 for t in txns if t["biz_type"] == "withdraw") == 2
+
+
 def test_withdraw_ambiguous_timeout_then_success_no_refund(client, monkeypatch) -> None:
     """#3 转账调用超时(异常),但查单确认已 SUCCESS → 不退款,单置 success。"""
     monkeypatch.setattr("app.integrations.wxpay.code_to_userinfo", lambda code: {"openid": "openid_amb", "nickname": None, "avatar_url": None, "raw": {}})