智能推荐改走离线库筛佣金率≥3%并隐藏距离;距离最近改实时搜索按用户真实位置排序
- 距离最近(/feed tab=distance):实时搜索召回(外卖搜"外卖" + 到店搜"美食"bizLine=1),都 sortField=6 离我最近,searchId 续页分页,用 App 上报的真实经纬度算距离; - 智能推荐(/feed tab=rec):从离线库 meituan_coupon 筛佣金率≥3% 分页返回(实测≥3% 的券几乎全是外卖、到店团购普遍<3%);并隐藏其距离展示(库里距离是相对城市默认点的、对用户无意义,置空后前端店名自动顶到最左); - feed 接口补 db 依赖与 _call/aliased/nullslast 导入。 Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
This commit is contained in:
chenshuobo
+84
-25
@@ -8,12 +8,12 @@ import logging
|
||||
from concurrent.futures import ThreadPoolExecutor, as_completed
|
||||
|
||||
from fastapi import APIRouter, Depends, HTTPException
|
||||
from sqlalchemy import select
|
||||
from sqlalchemy import nullslast, select
|
||||
from sqlalchemy.orm import Session, aliased
|
||||
|
||||
from app.core.config import settings
|
||||
from app.db.session import get_db
|
||||
from app.integrations.meituan import MeituanCpsError, get_referral_link, query_coupon
|
||||
from app.integrations.meituan import MeituanCpsError, _call, get_referral_link, query_coupon
|
||||
from app.models.meituan_coupon import MeituanCoupon
|
||||
from app.schemas.meituan import (
|
||||
CouponCard,
|
||||
@@ -93,7 +93,7 @@ def _commission_pct(card: CouponCard) -> float:
|
||||
|
||||
|
||||
@router.post("/feed", response_model=FeedResponse, summary="混合feed(外卖+到店交叉);tab=rec智能推荐/distance距离最近")
|
||||
def feed(req: FeedRequest) -> FeedResponse:
|
||||
def feed(req: FeedRequest, db: Session = Depends(get_db)) -> FeedResponse:
|
||||
if not settings.mt_cps_configured:
|
||||
return FeedResponse(items=[], has_next=False, page=req.page)
|
||||
lon, lat = req.longitude, req.latitude
|
||||
@@ -110,28 +110,87 @@ def feed(req: FeedRequest) -> FeedResponse:
|
||||
except MeituanCpsError:
|
||||
return []
|
||||
|
||||
# 智能推荐 / 距离最近:拉齐全部榜单轮次(~60 条),后端处理后【一次性返回、不分页】。
|
||||
# (池子很小;逐页实时打美团反而让客户端翻页卡顿/滑不动——一次拉齐 → 前端加载一次、滚动顺畅。)
|
||||
if tab in ("rec", "distance"):
|
||||
wm_all: list[dict] = []
|
||||
dd_all: list[dict] = []
|
||||
with ThreadPoolExecutor(max_workers=len(_TOPIC_ROUNDS) * 2) as pool:
|
||||
futs = [
|
||||
(pool.submit(_fetch_topic, 1, None, wm_topic),
|
||||
pool.submit(_fetch_topic, 2, 1, dd_topic))
|
||||
for wm_topic, dd_topic in _TOPIC_ROUNDS
|
||||
]
|
||||
for fw, fd in futs:
|
||||
wm_all += fw.result()
|
||||
dd_all += fd.result()
|
||||
cards = _interleave(wm_all, dd_all) # from_raw + 去重 + 外卖:到店 2:1 交叉
|
||||
if tab == "rec":
|
||||
# 智能推荐:去掉佣金率 < 3%
|
||||
cards = [c for c in cards if _commission_pct(c) >= 3.0]
|
||||
else:
|
||||
# 距离最近:在完整池上全局按距离由近及远(无距离排最后)
|
||||
cards.sort(key=lambda c: c.distance_meters if c.distance_meters is not None else float("inf"))
|
||||
return FeedResponse(items=cards, has_next=False, page=1)
|
||||
# 距离最近:搜索召回(外卖搜"外卖" + 到店搜"美食",都 sortField=6 离我最近)一页页拉。
|
||||
# 搜索翻页必须用 searchId(pageNo 翻不动),所以每个 feed 页顺序翻到第 N 页;两路并行、page 1 最快。
|
||||
# 无状态、不改 APP(传页码即可);按你位置实时算距离(库里没存 POI 经纬度,只能实时)。
|
||||
if tab == "distance":
|
||||
lon_i, lat_i = int(lon * 1_000_000), int(lat * 1_000_000)
|
||||
|
||||
def _search_page_n(platform: int, biz_line: int | None, keyword: str, n: int) -> tuple[list[dict], bool]:
|
||||
"""顺序翻到第 n 页(搜索须 searchId 续页),返回(第 n 页 items, 是否还有下一页)。"""
|
||||
sid: str | None = None
|
||||
data: list[dict] = []
|
||||
has_next = False
|
||||
for pg in range(1, n + 1):
|
||||
body: dict = {
|
||||
"platform": platform, "searchText": keyword, "sortField": 6,
|
||||
"longitude": lon_i, "latitude": lat_i, "pageSize": 20,
|
||||
}
|
||||
if biz_line is not None:
|
||||
body["bizLine"] = biz_line
|
||||
if sid:
|
||||
body["searchId"] = sid
|
||||
else:
|
||||
body["pageNo"] = 1
|
||||
try:
|
||||
r = _call("/cps_open/common/api/v1/query_coupon", body)
|
||||
except MeituanCpsError:
|
||||
return [], False
|
||||
data = r.get("data") or []
|
||||
sid = r.get("searchId")
|
||||
has_next = bool(r.get("hasNext")) and bool(data)
|
||||
if not data or (not has_next and pg < n):
|
||||
return [], False # 没那么多页了
|
||||
return data, has_next
|
||||
|
||||
with ThreadPoolExecutor(max_workers=2) as pool:
|
||||
f_wm = pool.submit(_search_page_n, 1, None, "外卖", req.page)
|
||||
f_dd = pool.submit(_search_page_n, 2, 1, "美食", req.page)
|
||||
(wm_data, wm_hn), (dd_data, dd_hn) = f_wm.result(), f_dd.result()
|
||||
|
||||
seen: set[str] = set()
|
||||
cards: list[CouponCard] = []
|
||||
for it in wm_data + dd_data:
|
||||
card = CouponCard.from_raw(it)
|
||||
if card.product_view_sign and card.product_view_sign not in seen:
|
||||
seen.add(card.product_view_sign)
|
||||
cards.append(card)
|
||||
cards.sort(key=lambda c: c.distance_meters if c.distance_meters is not None else float("inf"))
|
||||
return FeedResponse(items=cards, has_next=wm_hn or dd_hn, page=req.page)
|
||||
|
||||
# 智能推荐(rec):走【离线库】筛佣金率 ≥ 3%,分页返回(SQL 侧去重+排序+分页,秒级、不打美团)。
|
||||
# 实测库里佣金≥3% 去重后仅 ~578 条(几乎全是外卖;到店团购佣金普遍 <3%):实时按"同城热销榜单"
|
||||
# 拉既撞限流、又填不满(该榜单中位佣金 ~0.8%,筛完每页剩 0-1 条),故从库出。佣金阈值逻辑不变。
|
||||
if tab == "rec":
|
||||
PAGE = 20
|
||||
base = select(MeituanCoupon).where(MeituanCoupon.commission_percent >= 3.0)
|
||||
deduped = base.distinct(MeituanCoupon.dedup_key).order_by(
|
||||
MeituanCoupon.dedup_key,
|
||||
MeituanCoupon.commission_percent.desc(),
|
||||
).subquery()
|
||||
m = aliased(MeituanCoupon, deduped)
|
||||
start = (req.page - 1) * PAGE
|
||||
rows = db.execute(
|
||||
select(m)
|
||||
# 销量高的优先(无销量档排后),同档佣金高优先,id 兜底稳定分页
|
||||
.order_by(nullslast(m.sale_volume_num.desc()), m.commission_percent.desc(), m.id)
|
||||
.offset(start)
|
||||
.limit(PAGE + 1)
|
||||
).scalars().all()
|
||||
has_next = len(rows) > PAGE
|
||||
cards: list[CouponCard] = []
|
||||
for row in rows[:PAGE]:
|
||||
try:
|
||||
card = CouponCard.from_raw(row.raw or {})
|
||||
except Exception: # noqa: BLE001
|
||||
continue
|
||||
if card.product_view_sign:
|
||||
# 智能推荐不显示距离:库里的距离是相对城市默认点的(对用户无意义、且误导)。
|
||||
# 置空后前端"距离 店名"那行只剩店名、自动顶到最左(店名移到原距离的位置)。
|
||||
card.distance_text = None
|
||||
card.distance_meters = None
|
||||
cards.append(card)
|
||||
return FeedResponse(items=cards, has_next=has_next, page=req.page)
|
||||
|
||||
# 默认(老客户端不传 tab):沿用逐轮分页的混合 feed,不筛。
|
||||
page_idx = req.page - 1
|
||||
|
||||
Reference in New Issue
Block a user