feat: DAMA数据治理规则检查(财务七规则)— 后端/check-governance + 前端数据质量中心升级

后端 backend/app/api/data_quality.py:
- 新增 GET /api/cma/data-quality/check-governance?entity_id= 端点
- 7条规则: unit_check单位校验/dup_alert重复预警/orphan_check孤儿预警/virtual_pollution虚拟污染/entity_check实体归属/kpi_completeness KPI完整性/reconciliation勾稽验证
- 质量评分: error扣10/warning扣5, 每条规则最多扣一次, 最低0分
- 兼容实体筛选(0全部/1酣客/2博海), 向后兼容既有/check端点

前端 frontend/src/api/index.ts + views/DataQuality.vue:
- dataQualityApi.checkGovernance() 新增
- 页面新增治理规则区: 质量评分仪表盘 + 7规则红黄绿状态表 + 异常明细展开 + 重新检查按钮 + 实体筛选
This commit is contained in:
Hermes CI Fix
2026-08-30 07:26:59 +08:00
parent 240a9fa899
commit c36ed35348
3 changed files with 325 additions and 1 deletions
+195 -1
View File
@@ -3,7 +3,7 @@
"""
from fastapi import APIRouter, Depends, HTTPException, Query
from sqlalchemy.orm import Session
from sqlalchemy import func, and_
from sqlalchemy import func, and_, text
from typing import Optional
from datetime import datetime, timedelta
import json
@@ -322,3 +322,197 @@ def quality_stats(db: Session = Depends(get_db)):
"threshold_days": 180,
},
}
# ============================================================
# DAMA 数据治理规则检查(财务七规则)— 2026-08-30 P1
# ============================================================
RULES_META = {
"unit_check": {"name": "单位校验", "level": "error", "desc": "cash_plans.amount > 10000(万元口径可疑,疑似单位错乱)"},
"dup_alert": {"name": "重复预警", "level": "error", "desc": "同一plan_id存在多条pending应收预警(去重键错误)"},
"orphan_check": {"name": "孤儿预警", "level": "error", "desc": "预警suggestion.plan_id指向不存在的cash_plans记录"},
"virtual_pollution": {"name": "虚拟污染", "level": "error", "desc": "cash_plans.source含test/虚拟等测试标识混入真实数据"},
"entity_check": {"name": "实体归属", "level": "error", "desc": "kpi_values.entity_id与kpi_definitions.entity_id不一致"},
"kpi_completeness": {"name": "KPI完整性", "level": "warning", "desc": "active状态KPI无任何实际值的数量"},
"reconciliation": {"name": "勾稽验证", "level": "warning", "desc": "预算月度合计 vs 年度目标差异>20%"},
}
DETAIL_LIMIT = 10 # 每条规则detail最多列出的条数(避免响应过大)
def _run_rule_checks(db: Session, entity_id: int = 0):
"""执行7条DAMA治理规则,返回 issues 列表。entity_id=0 表示全部实体。"""
entity_filter = " AND cp.entity_id = :eid" if entity_id else ""
issues = []
# ── 规则1 单位校验 ──
rows = db.execute(text(
"SELECT cp.id, cp.entity_id, cp.amount, cp.source, cp.description "
"FROM cash_plans cp WHERE cp.amount > 10000" + entity_filter + " ORDER BY cp.amount DESC"
), {"eid": entity_id}).fetchall()
issues.append({
"rule": "unit_check", "level": "error",
"count": len(rows),
"detail": [f"plan#{r.id} 金额{r.amount}(疑似元)" for r in rows[:DETAIL_LIMIT]]
+ (["…等%d" % len(rows)] if len(rows) > DETAIL_LIMIT else []),
})
# ── 规则2 重复预警(同plan_id多条pending应收预警)──
if entity_id:
dup_sql = text(
"SELECT JSON_EXTRACT(a.suggestion, '$.plan_id') AS pid, COUNT(*) c, MAX(p.entity_id) eid "
"FROM kpi_alerts a JOIN cash_plans p ON p.id = JSON_EXTRACT(a.suggestion, '$.plan_id') "
"WHERE a.alert_type='cash_plan' AND a.status='pending' AND JSON_VALID(a.suggestion) "
"AND a.suggestion LIKE '%plan_id%' AND p.entity_id = :eid "
"GROUP BY pid HAVING c > 1 ORDER BY c DESC"
)
else:
dup_sql = text(
"SELECT JSON_EXTRACT(a.suggestion, '$.plan_id') AS pid, COUNT(*) c, MAX(p.entity_id) eid "
"FROM kpi_alerts a JOIN cash_plans p ON p.id = JSON_EXTRACT(a.suggestion, '$.plan_id') "
"WHERE a.alert_type='cash_plan' AND a.status='pending' AND JSON_VALID(a.suggestion) "
"AND a.suggestion LIKE '%plan_id%' "
"GROUP BY pid HAVING c > 1 ORDER BY c DESC"
)
dup_rows = db.execute(dup_sql, {"eid": entity_id}).fetchall()
issues.append({
"rule": "dup_alert", "level": "error",
"count": len(dup_rows),
"detail": [f"plan#{r.pid} 重复预警×{r.c}" for r in dup_rows[:DETAIL_LIMIT]]
+ (["…等%d个plan" % len(dup_rows)] if len(dup_rows) > DETAIL_LIMIT else []),
})
# ── 规则3 孤儿预警(plan_id指向不存在的cash_plans)──
orphan_sql = text(
"SELECT a.id, a.kpi_id, JSON_EXTRACT(a.suggestion, '$.plan_id') AS pid "
"FROM kpi_alerts a "
"WHERE a.alert_type='cash_plan' AND a.status='pending' AND JSON_VALID(a.suggestion) "
"AND a.suggestion LIKE '%plan_id%' "
"AND NOT EXISTS (SELECT 1 FROM cash_plans p WHERE p.id = JSON_EXTRACT(a.suggestion, '$.plan_id')) "
"ORDER BY a.id LIMIT 200"
)
orphan_rows = db.execute(orphan_sql).fetchall()
issues.append({
"rule": "orphan_check", "level": "error",
"count": len(orphan_rows),
"detail": [f"预警#{r.id}(kpi#{r.kpi_id}) → plan#{r.pid} 不存在" for r in orphan_rows[:DETAIL_LIMIT]]
+ (["…等%d" % len(orphan_rows)] if len(orphan_rows) > DETAIL_LIMIT else []),
})
# ── 规则4 虚拟污染(source含test/虚拟标识)──
rows = db.execute(text(
"SELECT cp.id, cp.entity_id, cp.source, cp.description FROM cash_plans cp "
"WHERE cp.source LIKE '%test%' OR cp.source LIKE '%虚拟%' OR cp.source LIKE '%demo%'"
+ entity_filter + " ORDER BY cp.id LIMIT 200"
), {"eid": entity_id}).fetchall()
issues.append({
"rule": "virtual_pollution", "level": "error",
"count": len(rows),
"detail": [f"plan#{r.id} source={r.source}" for r in rows[:DETAIL_LIMIT]]
+ (["…等%d" % len(rows)] if len(rows) > DETAIL_LIMIT else []),
})
# ── 规则5 实体归属(kpi_values.entity_id != kpi_definitions.entity_id)──
if entity_id:
ent_sql = text(
"SELECT v.id, v.kpi_id, d.kpi_code, v.entity_id AS v_eid, d.entity_id AS d_eid "
"FROM kpi_values v JOIN kpi_definitions d ON v.kpi_id = d.id "
"WHERE v.entity_id != d.entity_id AND v.entity_id = :eid ORDER BY v.id LIMIT 200"
)
else:
ent_sql = text(
"SELECT v.id, v.kpi_id, d.kpi_code, v.entity_id AS v_eid, d.entity_id AS d_eid "
"FROM kpi_values v JOIN kpi_definitions d ON v.kpi_id = d.id "
"WHERE v.entity_id != d.entity_id ORDER BY v.id LIMIT 200"
)
ent_rows = db.execute(ent_sql, {"eid": entity_id}).fetchall()
issues.append({
"rule": "entity_check", "level": "error",
"count": len(ent_rows),
"detail": [f"值#{r.id} {r.kpi_code} 实体{r.v_eid}≠定义实体{r.d_eid}" for r in ent_rows[:DETAIL_LIMIT]]
+ (["…等%d" % len(ent_rows)] if len(ent_rows) > DETAIL_LIMIT else []),
})
# ── 规则6 KPI完整性(active KPI无任何值)──
if entity_id:
comp_sql = text(
"SELECT d.id, d.kpi_code, d.kpi_name FROM kpi_definitions d "
"WHERE d.status='active' AND d.entity_id = :eid "
"AND NOT EXISTS (SELECT 1 FROM kpi_values v WHERE v.kpi_id = d.id) ORDER BY d.id LIMIT 300"
)
else:
comp_sql = text(
"SELECT d.id, d.kpi_code, d.kpi_name FROM kpi_definitions d "
"WHERE d.status='active' "
"AND NOT EXISTS (SELECT 1 FROM kpi_values v WHERE v.kpi_id = d.id) ORDER BY d.id LIMIT 300"
)
comp_rows = db.execute(comp_sql, {"eid": entity_id}).fetchall()
issues.append({
"rule": "kpi_completeness", "level": "warning",
"count": len(comp_rows),
"detail": [f"{r.kpi_code} {r.kpi_name}(无值)" for r in comp_rows[:DETAIL_LIMIT]]
+ (["…等%d个KPI" % len(comp_rows)] if len(comp_rows) > DETAIL_LIMIT else []),
})
# ── 规则7 勾稽验证(预算月度合计 vs 年度目标差异>20%)──
if entity_id:
recon_sql = text(
"SELECT d.kpi_code, d.kpi_name, d.target_yearly, "
"SUM(b.budget_value) AS monthly_sum, "
"ROUND((SUM(b.budget_value) - d.target_yearly) / d.target_yearly * 100, 1) AS diff_pct "
"FROM kpi_definitions d JOIN budget_plans b ON b.kpi_id = d.id "
"WHERE d.status='active' AND d.target_yearly > 0 AND d.entity_id = :eid "
"GROUP BY d.id HAVING ABS(diff_pct) > 20 ORDER BY ABS(diff_pct) DESC LIMIT 200"
)
else:
recon_sql = text(
"SELECT d.kpi_code, d.kpi_name, d.target_yearly, "
"SUM(b.budget_value) AS monthly_sum, "
"ROUND((SUM(b.budget_value) - d.target_yearly) / d.target_yearly * 100, 1) AS diff_pct "
"FROM kpi_definitions d JOIN budget_plans b ON b.kpi_id = d.id "
"WHERE d.status='active' AND d.target_yearly > 0 "
"GROUP BY d.id HAVING ABS(diff_pct) > 20 ORDER BY ABS(diff_pct) DESC LIMIT 200"
)
recon_rows = db.execute(recon_sql, {"eid": entity_id}).fetchall()
issues.append({
"rule": "reconciliation", "level": "warning",
"count": len(recon_rows),
"detail": [f"{r.kpi_code} 预算合计{round(r.monthly_sum, 1)} vs 年度目标{r.target_yearly} 差异{r.diff_pct}%" for r in recon_rows[:DETAIL_LIMIT]]
+ (["…等%d个KPI" % len(recon_rows)] if len(recon_rows) > DETAIL_LIMIT else []),
})
return issues
@router.get("/check-governance")
def check_governance(
entity_id: Optional[int] = Query(0, description="实体ID: 0=全部, 1=酣客, 2=博海"),
db: Session = Depends(get_db),
):
"""DAMA数据治理规则检查(财务七规则)— 返回质量评分+异常清单。
评分规则: 满分100,error级规则每条扣10分,warning级规则每条扣5分,
每条规则最多扣一次分(按规则是否命中,不按count累扣),最低0分。
"""
issues = _run_rule_checks(db, entity_id or 0)
# 计算评分
score = 100
for item in issues:
if item["count"] > 0:
score -= 10 if item["level"] == "error" else 5
score = max(0, score)
passed = [item["rule"] for item in issues if item["count"] == 0]
return {
"checked_at": datetime.now().isoformat(timespec="seconds"),
"entity_id": entity_id or 0,
"score": score,
"total_rules": len(issues),
"issues": issues,
"passed": passed,
"rules_meta": RULES_META,
}