init: Prefect ETL sync project (RIIN/SDS → PG) with 12 sync flows

This commit is contained in:
yeuimu
2026-07-23 08:13:09 +00:00
commit 4cba1da4e1
41 changed files with 6219 additions and 0 deletions
+56
View File
@@ -0,0 +1,56 @@
"""RIIN 近期同步:拉取最近 N 天订单(默认 7 天,1~90)。
对应 riin-recent-sync。from = startOfDay(now - N days),to = now。
"""
from __future__ import annotations
from datetime import datetime, timedelta
from prefect import flow, task
from flows.riin._pipeline import sync_riin_records
from flows.shared import (
SyncDB, create_riin_client, fetch_riin_all_pages, format_date, start_of_day,
)
@task(retries=3, retry_delay_seconds=[3, 6, 9])
def extract_orders(params: dict) -> list:
client = create_riin_client()
try:
return fetch_riin_all_pages(client, params)
finally:
client.close()
@task
def upsert_records(raw_orders: list) -> dict:
client = create_riin_client()
try:
return sync_riin_records(raw_orders, client, SyncDB())
finally:
client.close()
@flow(name="riin-recent-sync", log_prints=True)
def riin_recent_sync(recent_days: int = 7) -> dict:
days = min(max(recent_days, 1), 90)
now = datetime.now()
from_dt = start_of_day(now - timedelta(days=days))
params = {
"time_type": 3,
"start_time": format_date(from_dt),
"end_time": format_date(now),
"page_size": 200,
}
print(f"RIIN 近期同步 近 {days} 天 {params['start_time']} ~ {params['end_time']}")
raw = extract_orders(params)
stats = upsert_records(raw)
print(f"完成: 订单 {stats['orderCount']} / 商品 {stats['goodsCount']}")
return {"syncRange": {"from": params["start_time"], "to": params["end_time"]}, "recentDays": days, **stats}
if __name__ == "__main__":
riin_recent_sync.serve(name="riin-recent-sync", cron="30 * * * *")