lzf_0626
df96275ab5
实测确认 #21 描述准确：fin_risk_alert 只有 11 个普通 BTREE 索引 + 主键 + alert_no 唯一键，规则命中的 JSON_CONTAINS 查询 EXPLAIN 为 type=ALL、possible_keys=NULL， 即全表扫。MySQL 8.0.27 支持多值索引，故新增迁移 20260911_risk_rule_index： ADD INDEX idx_fin_risk_alert_trigger_rule_codes ((CAST(`trigger_rule_codes` AS CHAR(16) ARRAY))) 迁移幂等（先查 information_schema.STATISTICS），upgrade/downgrade 往返已验证。 生效后 EXPLAIN 变为 access_type=range 且 key 命中该索引，原始证据留档在 docs/evidence/risk-index-probe.json（由 tools/probe_risk_index.py 生成，只读探查）。 只解决一半，另一半如实记为限制：若干 like(f"%{keyword}%") 全表扫无法用 B-tree 索引， 根治需全文索引 + 中文分词组件（部署依赖），本轮不做。 revision 名刻意压到 32 字符以内 —— alembic_version.version_num 是 VARCHAR(32)， 超长会在写版本号时报 1406，而 DDL 是非事务的，那时索引已经建好了。 docs/25 追加"P3 处理结果"表，逐条登记 17-25 的状态：#19 是协议级重做（keyset 分页） 不单方面改，#21 部分修复，#25 前半段不成立，其余已修。
139 lines
4.4 KiB
Python
139 lines
4.4 KiB
Python
"""只读探查：`fin_risk_alert.trigger_rule_codes` 能否用 JSON 多值索引（docs/25 P3 #21）。
|
||
|
||
只做 SELECT / EXPLAIN，不做任何写入。结果写成 JSON 便于在 GBK 控制台下查看：
|
||
|
||
python tools/probe_risk_index.py
|
||
|
||
排查点：
|
||
1. MySQL 版本是否 ≥ 8.0.17（多值索引的下限）；
|
||
2. 列是否确实是 JSON、是否已存在同类索引；
|
||
3. 是否存在**非数组**取值 —— 有的话 `CAST(... AS CHAR ARRAY)` 建索引会直接失败；
|
||
4. 当前 `JSON_CONTAINS` 查询走的是什么访问路径（有没有可用的 key）。
|
||
"""
|
||
|
||
from __future__ import annotations
|
||
|
||
import asyncio
|
||
import json
|
||
from pathlib import Path
|
||
from typing import Any
|
||
|
||
from sqlalchemy import text
|
||
|
||
from app.infrastructure.db import SessionFactory
|
||
|
||
OUTPUT = Path("docs/evidence/risk-index-probe.json")
|
||
|
||
|
||
async def collect() -> dict[str, Any]:
|
||
report: dict[str, Any] = {}
|
||
async with SessionFactory() as session:
|
||
report["mysql_version"] = (
|
||
await session.execute(text("SELECT VERSION()"))
|
||
).scalar_one()
|
||
|
||
report["column"] = [
|
||
dict(row)
|
||
for row in (
|
||
(
|
||
await session.execute(
|
||
text(
|
||
"""
|
||
SELECT COLUMN_NAME, COLUMN_TYPE, DATA_TYPE, IS_NULLABLE
|
||
FROM information_schema.COLUMNS
|
||
WHERE TABLE_SCHEMA = DATABASE()
|
||
AND TABLE_NAME = 'fin_risk_alert'
|
||
AND COLUMN_NAME = 'trigger_rule_codes'
|
||
"""
|
||
)
|
||
)
|
||
)
|
||
.mappings()
|
||
.all()
|
||
)
|
||
]
|
||
|
||
report["indexes"] = [
|
||
dict(row)
|
||
for row in (
|
||
(
|
||
await session.execute(
|
||
text(
|
||
"""
|
||
SELECT INDEX_NAME, COLUMN_NAME, INDEX_TYPE, EXPRESSION
|
||
FROM information_schema.STATISTICS
|
||
WHERE TABLE_SCHEMA = DATABASE()
|
||
AND TABLE_NAME = 'fin_risk_alert'
|
||
ORDER BY INDEX_NAME, SEQ_IN_INDEX
|
||
"""
|
||
)
|
||
)
|
||
)
|
||
.mappings()
|
||
.all()
|
||
)
|
||
]
|
||
|
||
report["row_count"] = (
|
||
await session.execute(text("SELECT COUNT(*) FROM fin_risk_alert"))
|
||
).scalar_one()
|
||
report["non_array_rows"] = (
|
||
await session.execute(
|
||
text(
|
||
"SELECT COUNT(*) FROM fin_risk_alert "
|
||
"WHERE trigger_rule_codes IS NULL "
|
||
" OR JSON_TYPE(trigger_rule_codes) <> 'ARRAY'"
|
||
)
|
||
)
|
||
).scalar_one()
|
||
report["distinct_rule_codes"] = [
|
||
dict(row)
|
||
for row in (
|
||
(
|
||
await session.execute(
|
||
text(
|
||
"""
|
||
SELECT CAST(trigger_rule_codes AS CHAR) AS raw_codes,
|
||
COUNT(*) AS rows_count
|
||
FROM fin_risk_alert
|
||
GROUP BY 1
|
||
"""
|
||
)
|
||
)
|
||
)
|
||
.mappings()
|
||
.all()
|
||
)
|
||
]
|
||
|
||
# 用 FORMAT=JSON：传统 EXPLAIN 会把"该查询无法被缓存"写成 warning 打到 stderr，
|
||
# 让脚本在 CI 里看起来是失败的，而它其实成功了。
|
||
raw_explain = (
|
||
await session.execute(
|
||
text(
|
||
"""
|
||
EXPLAIN FORMAT=JSON
|
||
SELECT id FROM fin_risk_alert
|
||
WHERE JSON_CONTAINS(trigger_rule_codes, '"RW-018"')
|
||
"""
|
||
)
|
||
)
|
||
).scalar_one()
|
||
report["explain_json_contains"] = json.loads(raw_explain)
|
||
|
||
return report
|
||
|
||
|
||
async def main() -> None:
|
||
report = await collect()
|
||
OUTPUT.parent.mkdir(parents=True, exist_ok=True)
|
||
OUTPUT.write_text(
|
||
json.dumps(report, ensure_ascii=False, indent=2, default=str),
|
||
encoding="utf-8",
|
||
)
|
||
print(f"wrote {OUTPUT}")
|
||
|
||
|
||
if __name__ == "__main__":
|
||
asyncio.run(main())
|