Merge pull request 'fix: 写库重试 3→5 次、退避拉长至 1/3/7/15/30s' (#9) from inkreach/harden-db-write-retry into main

Reviewed-on: #9

.python-version

@@ -1 +0,0 @@
-3.12

api/ocr/models.py

@@ -37,16 +37,16 @@ class OcrResult(BaseModel):
 
 
 # OCR 服务已有分类器模型的国家(版面见 api.md §4)
-SUPPORTED_COUNTRIES = {"US", "UK", "DE", "MX", "SA", "CA", "ES", "IT", "PL"}
+SUPPORTED_COUNTRIES = {"US", "UK", "DE", "MX", "SA", "CA", "ES", "IT"}
 
 # ISO 国家代码 / 中文国名 → OCR 国家代码
 _COUNTRY_MAP: dict[str, str] = {
     # 代码
     "GB": "UK", "UK": "UK", "US": "US", "DE": "DE", "MX": "MX", "SA": "SA",
-    "CA": "CA", "ES": "ES", "IT": "IT", "PL": "PL",
+    "CA": "CA", "ES": "ES", "IT": "IT",
     # 中文名
     "英国": "UK", "美国": "US", "德国": "DE", "墨西哥": "MX", "沙特": "SA",
-    "加拿大": "CA", "西班牙": "ES", "意大利": "IT", "波兰": "PL",
+    "加拿大": "CA", "西班牙": "ES", "意大利": "IT",
 }
 
 

flows/label_ocr.py

@@ -29,20 +29,13 @@ _OCR_FIELD_MAP = [
 ]
 
 
-# 单号不需要 "-" 的承运商(Evri 单号两段式带 "-",下游查询用无 "-" 形式)
-_DASH_STRIP_COURIERS = {"EVRI"}
-
-
 def _result_to_fields(r) -> dict[str, str | None]:
     if not r.ok:
         err = (r.error or "").lower()
         if "reject" in err or "not a recognized layout" in err:
             return {"ocrLayout": "UNSUPPORTED"}
         return {"ocrLayout": "ERROR"}
-    fields = {col: (getattr(r, attr) or "") for attr, col in _OCR_FIELD_MAP}
-    if (r.courier or "").strip().upper() in _DASH_STRIP_COURIERS:
-        fields["ocrTracking"] = fields["ocrTracking"].replace("-", "")
-    return fields
+    return {col: (getattr(r, attr) or "") for attr, col in _OCR_FIELD_MAP}
 
 
 def _has_running_peer() -> bool:
@@ -149,7 +142,7 @@ def label_ocr(
     手动跑指定国家: label_ocr(country_regions=['德国'])
     """
     if country_regions is None:
-        country_regions = ['英国', '美国', '墨西哥', '德国', '沙特', '加拿大', '西班牙', '意大利', '波兰']
+        country_regions = ['英国', '美国', '墨西哥', '德国', '沙特', '加拿大', '西班牙', '意大利']
     if _has_running_peer():
         print("已有 label-ocr 在运行,跳过本次")
         return {"skipped": True}

flows/platform_order/sync_shop_dim.py

@@ -1,144 +0,0 @@
-# -*- coding: utf-8 -*-
-"""shop_dim 同步 flow：teable(源) -> 本地 smt 库(目标)。
-
-最小包装自 sync_shop_dim.py,对应 shop-dim-sync,每 5 分钟调度一次。
-源: PostgreSQL @ table.inkreach.cc:10432  db=teable  schema=bsehWCxpH6HoHUdkjUd
-    物理表 tblUF7OrBxeRKUyWBc2 (teable 里名为 "shop_dim" 的表)
-目标: PostgreSQL @ 192.168.124.92:5432  db=smt  public.shop_dim
-
-本地直接跑:  python flows/platform_order/sync_shop_dim.py
-"""
-
-import time
-from typing import Any
-
-import psycopg2
-from prefect import flow
-
-SRC = dict(host="table.inkreach.cc", port=10432, dbname="teable",
-           user="read_only_role_bsehWCxpH6HoHUdkjUd",
-           password="ygBuh8Au5J-PjuNQASos0", connect_timeout=25)
-SRC_TABLE = '"bsehWCxpH6HoHUdkjUd"."tblUF7OrBxeRKUyWBc2"'
-
-DST = dict(host="192.168.124.92", port=5432, dbname="smt",
-           user="postgres", password="Inkreach123456.", connect_timeout=15)
-
-COL_MAP = [
-    ("Guo_Jia",            "country"),
-    ("Dian_Pu_Ming_Cheng", "shop_name"),
-    ("DBDian_Pu_Ming",     "db_store_name"),
-    ("Ping_Tai",           "platform"),
-    ("Lei_Xing",           "shop_type"),
-    ("Deng_Ji",            "tier"),
-    ("Fu_Ze_Ren",          "owner"),
-    ("Pin_Lei",            "category"),
-    ("Xia_Dian_Shi_Jian",  "opened_at"),
-    ("Mu_Biao_Jian_Shu",   "sep_target_qty"),
-]
-
-
-def log_msg(msg):
-    print(msg)
-
-
-def connect(cfg, tries=5):
-    last = None
-    for i in range(tries):
-        try:
-            return psycopg2.connect(**cfg)
-        except Exception as e:
-            last = e
-            log_msg(f"  连接重试 {i+1}/{tries}: {e}")
-            time.sleep(3)
-    raise ConnectionError(f"连接失败: {last}")
-
-
-def norm(v, col):
-    if v is None:
-        return None
-    if col == "opened_at":
-        return v.date() if hasattr(v, "date") else v
-    if col == "sep_target_qty":
-        try:
-            return int(round(float(v)))
-        except Exception:
-            return None
-    s = str(v).strip()
-    return s if s != "" else None
-
-
-@flow(name="shop-dim-sync", log_prints=True)
-def shop_dim_sync(dry_run: bool = False, prune: bool = True) -> dict[str, Any]:
-    log_msg(f"[1/4] 读取源表 {SRC_TABLE} ...")
-    sc = connect(SRC)
-    scur = sc.cursor()
-    sel = ", ".join('"%s"' % c for c, _ in COL_MAP)
-    scur.execute("SELECT %s FROM %s" % (sel, SRC_TABLE))
-    raw = scur.fetchall()
-    scur.close(); sc.close()
-    src_rows = [
-        dict(zip([d for _, d in COL_MAP], [norm(v, c) for v, (_, c) in zip(row, COL_MAP)]))
-        for row in raw
-    ]
-    log_msg(f"      源记录: {len(src_rows)} 行")
-
-    log_msg("[2/4] 连接目标 smt.shop_dim ...")
-    dc = connect(DST)
-    dcur = dc.cursor()
-    dcur.execute("SELECT country, db_store_name FROM shop_dim WHERE db_store_name IS NOT NULL")
-    existing = {(r[0], r[1]) for r in dcur.fetchall()}
-    log_msg(f"      目标现有: {len(existing)} 个 (country, db_store_name)")
-
-    src_keys = {(r["country"], r["db_store_name"]) for r in src_rows}
-    to_update = src_keys & existing
-    to_insert = src_keys - existing
-    to_delete = existing - src_keys
-    log_msg(f"[3/4] 计划: 更新 {len(to_update)} | 新增 {len(to_insert)} | 待删 {len(to_delete)}")
-
-    if dry_run:
-        log_msg("[dry-run] 不执行写入。")
-        if to_delete:
-            log_msg("        待删示例: " + str(list(to_delete)[:10]))
-        dcur.close(); dc.close()
-        return {"sourceRows": len(src_rows), "dryRun": True}
-
-    cols = [d for _, d in COL_MAP]
-    upd = 0
-    ins = 0
-    for r in src_rows:
-        vals = [r[c] for c in cols]
-        if (r["country"], r["db_store_name"]) in existing:
-            set_clause = ", ".join(f"{c}=%s" for c in cols)
-            dcur.execute(
-                f"UPDATE shop_dim SET {set_clause} WHERE country=%s AND db_store_name=%s",
-                vals + [r["country"], r["db_store_name"]])
-            upd += dcur.rowcount
-        else:
-            ph = ", ".join(["%s"] * len(cols))
-            dcur.execute(f"INSERT INTO shop_dim ({', '.join(cols)}) VALUES ({ph})", vals)
-            ins += dcur.rowcount
-
-    dele = 0
-    if prune and to_delete:
-        for country, dbn in to_delete:
-            dcur.execute("DELETE FROM shop_dim WHERE country=%s AND db_store_name=%s", (country, dbn))
-            dele += dcur.rowcount
-
-    dc.commit()
-    log_msg(f"[4/4] 完成: 更新 {upd} | 新增 {ins} | 删除 {dele}")
-    dcur.execute("SELECT COUNT(*) FROM shop_dim WHERE db_store_name IS NOT NULL")
-    total = dcur.fetchone()[0]
-    log_msg(f"      目标现有总数: {total}")
-    dcur.close(); dc.close()
-
-    return {
-        "sourceRows": len(src_rows),
-        "updated": upd,
-        "inserted": ins,
-        "deleted": dele,
-        "total": total,
-    }
-
-
-if __name__ == "__main__":
-    shop_dim_sync.serve(name="shop-dim-sync", cron="*/5 * * * *")

flows/shared/clients.py

@@ -8,8 +8,6 @@
 from __future__ import annotations
 
 import json
-import random
-import threading
 import time
 from datetime import datetime, timedelta, timezone
 from typing import Any, Callable, TypeVar
@@ -23,9 +21,7 @@ from flows.shared.mappers import map_sds_channel_type
 
 T = TypeVar("T")
 
-GOODS_CONCURRENCY = 5
-GOODS_MIN_INTERVAL_MS = 340  # 相邻两次 order/goods 调用全局最小间隔(峰值 < 3 次/秒)
-GOODS_MAX_JITTER_MS = 160  # 在最小间隔上叠加的随机抖动上限,实际间隔 340~500ms 随机
+GOODS_CONCURRENCY = 20
 PAGE_DELAY_MS = 500
 MAX_RETRIES = 3
 RETRY_DELAY_MS = 3000
@@ -148,25 +144,8 @@ def fetch_sds_all_pages(client: SdsMngClient, params: dict[str, Any]) -> list[An
 
 # ---------- RIIN 商品 ----------
 
-# 全局限速:RIIN 分销侧反馈 distribution/order/goods 逐单调用对其服务器压力过大,
-# 所有线程共享节拍,相邻两次调用间隔 = GOODS_MIN_INTERVAL_MS + 随机抖动(340~500ms,峰值 < 3 次/秒)。
-_goods_pace_lock = threading.Lock()
-_goods_last_call_at = 0.0
-
-
-def _pace_goods_call() -> None:
-    global _goods_last_call_at
-    with _goods_pace_lock:
-        interval = (GOODS_MIN_INTERVAL_MS + random.uniform(0, GOODS_MAX_JITTER_MS)) / 1000
-        wait = _goods_last_call_at + interval - time.monotonic()
-        if wait > 0:
-            time.sleep(wait)
-        _goods_last_call_at = time.monotonic()
-
-
 def fetch_goods_for_order(client: RiinDisClient, order_id: str) -> list[Any]:
     try:
-        _pace_goods_call()
         resp = client.get_order_goods(order_id)
         return list(resp.data)
     except Exception as err:  # noqa: BLE001

prefect.yaml

@@ -21,7 +21,6 @@ deployments:
   schedules:
   - cron: 30 * * * *
     timezone: Asia/Shanghai
-    active: false  # 暂停:每小时重拉近7天全部订单商品,对 RIIN distribution/order/goods 调用量过大(增量/刷新未完结/日报核验已覆盖)
   work_pool:
     name: inkreach-pool
 - name: riin-refresh-pending
@@ -40,7 +39,6 @@ deployments:
   schedules:
   - cron: 0 6 * * *
     timezone: Asia/Shanghai
-    active: false  # 暂停:与 daily-report 18:30 的核验子流程重复,保留日报一处即可
   work_pool:
     name: inkreach-pool
 - name: sds-order-sync
@@ -110,13 +108,6 @@ deployments:
     timezone: Asia/Shanghai
   work_pool:
     name: inkreach-pool
-- name: shop-dim-sync
-  entrypoint: flows/platform_order/sync_shop_dim.py:shop_dim_sync
-  schedules:
-  - cron: '*/5 * * * *'
-    timezone: Asia/Shanghai
-  work_pool:
-    name: inkreach-pool
 - name: label-ocr
   entrypoint: flows/label_ocr.py:label_ocr
   schedules: