hankin
423baff73b
- Docker bridge 网络隔离（8000 端口封死） - Gunicorn 4 Worker 多进程 - Alembic 数据库迁移基线 - 日志轮转 20m×3 - JWT 密钥 + DB 密码 + CORS 收紧 - 3-2-1 备份链路（NAS + R740-B 冷备） - 连接池 pool_pre_ping + pool_recycle=3600
128 lines
4.2 KiB
Python
128 lines
4.2 KiB
Python
# -*- coding: utf-8 -*-
|
|
"""
|
|
销售数据分析服务 (Dify BaaS 版本)
|
|
基于 SQL 预聚合的销售漏斗统计 + Dify AI 驱动的复盘报告生成。
|
|
"""
|
|
|
|
import logging
|
|
from datetime import datetime, timezone
|
|
|
|
from sqlalchemy import func, select, extract, case
|
|
from sqlalchemy.ext.asyncio import AsyncSession
|
|
|
|
from app.core.config import settings
|
|
from app.core.dify_client import dify_client
|
|
from app.models.crm_business import SalesOpportunity
|
|
|
|
logger = logging.getLogger("analytics")
|
|
|
|
|
|
# ============================================================
|
|
# 1. SQL 预聚合：销售漏斗指标
|
|
# ============================================================
|
|
|
|
async def get_sales_metrics(db: AsyncSession) -> list[dict]:
|
|
"""
|
|
按当月统计各销售阶段的机会数量和总金额。
|
|
使用 SQLAlchemy GROUP BY 聚合查询，在数据库层面完成计算。
|
|
"""
|
|
now = datetime.now(timezone.utc)
|
|
|
|
stmt = (
|
|
select(
|
|
SalesOpportunity.stage,
|
|
func.count(SalesOpportunity.id).label("count"),
|
|
func.coalesce(func.sum(SalesOpportunity.amount), 0).label("total_amount"),
|
|
)
|
|
.where(
|
|
extract("year", SalesOpportunity.created_at) == now.year,
|
|
extract("month", SalesOpportunity.created_at) == now.month,
|
|
)
|
|
.group_by(SalesOpportunity.stage)
|
|
.order_by(
|
|
case(
|
|
(SalesOpportunity.stage == "意向", 1),
|
|
(SalesOpportunity.stage == "谈判", 2),
|
|
(SalesOpportunity.stage == "成交", 3),
|
|
(SalesOpportunity.stage == "流失", 4),
|
|
else_=5,
|
|
)
|
|
)
|
|
)
|
|
|
|
result = await db.execute(stmt)
|
|
rows = result.all()
|
|
|
|
metrics = [
|
|
{
|
|
"stage": row.stage,
|
|
"count": row.count,
|
|
"total_amount": float(row.total_amount),
|
|
}
|
|
for row in rows
|
|
]
|
|
|
|
logger.info("当月销售指标: %s", metrics)
|
|
return metrics
|
|
|
|
|
|
# ============================================================
|
|
# 2. Dify AI 复盘报告生成
|
|
# ============================================================
|
|
|
|
async def generate_monthly_report(db: AsyncSession) -> dict:
|
|
"""
|
|
生成当月销售复盘报告：
|
|
1. SQL 预聚合获取真实数据指标
|
|
2. 将结构化数据通过 inputs 传给 Dify 报告 App
|
|
3. 直接返回 Dify 生成的报告文本
|
|
|
|
:return: {"metrics": [...], "report": "Dify 生成的报告文本"}
|
|
"""
|
|
# Step 1: 获取真实销售数据
|
|
metrics = await get_sales_metrics(db)
|
|
|
|
if not metrics:
|
|
return {
|
|
"metrics": [],
|
|
"report": "当月暂无销售机会数据，无法生成复盘报告。",
|
|
}
|
|
|
|
# Step 2: 将数据序列化为 Dify 可消费的文本格式
|
|
# *** inputs 的键名必须与 Dify 后台配置的变量名对齐 ***
|
|
# 在 Dify 后台的报告 App 编排中，需定义输入变量 "metrics_data"
|
|
metrics_text = "\n".join(
|
|
f"- {m['stage']}: {m['count']} 个机会, 总金额 ¥{m['total_amount']:,.2f}"
|
|
for m in metrics
|
|
)
|
|
total_count = sum(m["count"] for m in metrics)
|
|
total_amount = sum(m["total_amount"] for m in metrics)
|
|
metrics_text += f"\n- 合计: {total_count} 个机会, 总金额 ¥{total_amount:,.2f}"
|
|
|
|
# Step 3: 调用 Dify 报告 App
|
|
if not settings.DIFY_REPORT_APP_API_KEY:
|
|
logger.error("DIFY_REPORT_APP_API_KEY 未配置")
|
|
return {
|
|
"metrics": metrics,
|
|
"report": "AI 报告服务未配置，请联系管理员。",
|
|
}
|
|
|
|
# 动态生成 Dify 后台所需的必填参数 report_period
|
|
now = datetime.now(timezone.utc)
|
|
report_period = f"{now.year}年{now.month:02d}月"
|
|
|
|
report_text = await dify_client.call_text_generator(
|
|
api_key=settings.DIFY_REPORT_APP_API_KEY,
|
|
inputs={"metrics_data": metrics_text, "report_period": report_period},
|
|
)
|
|
|
|
if not report_text:
|
|
report_text = "AI 报告生成失败，请稍后重试或检查 Dify 服务状态。"
|
|
|
|
logger.info("月度复盘报告生成完成 (%d 字)", len(report_text))
|
|
|
|
return {
|
|
"metrics": metrics,
|
|
"report": report_text,
|
|
}
|