#!/usr/bin/env python3
# 在 ads-center 上运行：从 report_api.db 取 mintegral/bigo/mediago 三渠道
# 按 campid 匹配 Voypace 产品，聚合(日期+campid+国家)，输出 JSON 到 stdout。
# 参数：start_date（含）  end_date（含），格式 yyyy-MM-dd
import sqlite3, json, sys, os
from collections import defaultdict

DB = "/Users/ads-center/Documents/database/report_api.db"
START = sys.argv[1] if len(sys.argv) > 1 else "2026-07-15"
END   = sys.argv[2] if len(sys.argv) > 2 else "2026-12-31"

# campid -> 产品信息（从同目录 campaigns.json 读取）
HERE = os.path.dirname(os.path.abspath(__file__))
_camps = json.load(open(os.path.join(HERE, "campaigns.json")))
CAMPS = {r["campid"]: {"product": r["product"], "pkg": r["pkg"]} for r in _camps}

conn = sqlite3.connect(DB)
conn.row_factory = sqlite3.Row

# 聚合容器：key=(date,channel,campid,location) -> 累加指标
agg = defaultdict(lambda: {"impression":0,"clicks":0,"conversions":0,"spend":0.0})
meta = {}

def campid_of(text):
    t = str(text or "")
    for cid in CAMPS:
        if cid in t:
            return cid
    return None

# ---- Mintegral ----
# 广告单元名(如 91555309_Viu_ID)可能落在 Offer_Name 或 Campaign_Id，两字段都匹配
like = " OR ".join([f"Offer_Name LIKE '%{cid}%' OR Campaign_Id LIKE '%{cid}%'" for cid in CAMPS])
sql = f"""SELECT Date d, Offer_Name nm, Campaign_Id cmp, Location loc, Impression imp, Click clk,
                 Conversion cv, Spend sp
          FROM mintegral WHERE Date>=? AND Date<=? AND ({like})"""
for r in conn.execute(sql, (START, END)):
    cid = campid_of(r["nm"]) or campid_of(r["cmp"])
    if not cid: continue
    k = (r["d"], "Mintegral", cid, (r["loc"] or "").upper())
    a = agg[k]
    a["impression"] += int(r["imp"] or 0); a["clicks"] += int(r["clk"] or 0)
    a["conversions"] += int(r["cv"] or 0); a["spend"] += float(r["sp"] or 0)
    meta[k] = (r["nm"] or r["cmp"], CAMPS[cid])

# ---- Bigo ----
like = " OR ".join([f"campaignName LIKE '%{cid}%'" for cid in CAMPS])
sql = f"""SELECT aggregateTime d, campaignName nm, country loc, tgtPkgName pkg,
                 impression imp, click clk, real_result_conversion cv, totalCost cost
          FROM bigo WHERE aggregateTime>=? AND aggregateTime<=? AND ({like})"""
for r in conn.execute(sql, (START, END)):
    cid = campid_of(r["nm"])
    if not cid: continue
    k = (r["d"], "Bigo", cid, (r["loc"] or "").upper())
    a = agg[k]
    a["impression"] += int(r["imp"] or 0); a["clicks"] += int(r["clk"] or 0)
    a["conversions"] += int(r["cv"] or 0); a["spend"] += float(r["cost"] or 0)/100.0
    meta[k] = (r["nm"], CAMPS[cid])

# ---- MediaGo ---- (ad级明细，需聚合；spend = cpc*click 反算)
like = " OR ".join([f"campaign_name LIKE '%{cid}%'" for cid in CAMPS])
sql = f"""SELECT date d, campaign_name nm, adset_country loc, vimp imp, click clk,
                 cv cvn, cpc
          FROM mediago WHERE date>=? AND date<=? AND ({like})"""
for r in conn.execute(sql, (START, END)):
    cid = campid_of(r["nm"])
    if not cid: continue
    k = (r["d"], "MediaGo", cid, (r["loc"] or "").upper())
    a = agg[k]
    clk = int(r["clk"] or 0)
    a["impression"] += int(r["imp"] or 0); a["clicks"] += clk
    a["conversions"] += int(r["cvn"] or 0); a["spend"] += float(r["cpc"] or 0) * clk
    meta[k] = (r["nm"], CAMPS[cid])

conn.close()

out = []
for k, a in agg.items():
    date, channel, cid, loc = k
    nm, info = meta[k]
    imp, clk, cv = a["impression"], a["clicks"], a["conversions"]
    out.append({
        "date": date, "channel": channel, "campid": cid,
        "offer_name": nm, "product": info["product"], "pkg": info["pkg"],
        "location": loc, "bid_type": "",
        "impression": imp, "clicks": clk, "conversions": cv,
        "ctr": round(clk/imp, 5) if imp else 0.0,
        "cvr": round(cv/clk, 5) if clk else 0.0,
        "spend": round(a["spend"], 2),
    })
out.sort(key=lambda x: (x["date"], x["channel"], -x["impression"]))
print(json.dumps(out, ensure_ascii=False))