where条件中有权限校验的自定义函数优化方法
Posted robinson1988
tags:
篇首语：本文由小常识网(cha138.com)小编为大家整理，主要介绍了where条件中有权限校验的自定义函数优化方法相关的知识，希望对你有一定的参考价值。
下面SQL要跑47秒(SQL已做脱敏处理，最原始SQL要跑几十分钟，添加HINT修正执行计划后跑47秒)
54316 rows inserted
Executed in 47.031 seconds
INSERT INTO TMP
SELECT 1,
ORG_ID,
SESSION_ID,
NULL,
NULL,
NULL,
DEPARTMENT_CODE,
NULL,
DEPARTMENT_NAME,
NULL,
SALES_PERSON,
NULL,
PROJECT_NUMBER,
NULL,
NULL,
NULL,
NULL,
NULL,
NULL,
NULL,
ITEM_CATEGORY,
ITEM_CATEGORY_DESC,
NULL,
SO_TYPE,
SO_AREA,
NULL,
NULL,
NULL,
NULL,
NULL,
NULL,
SO_COST,
INTERNAL_SO_COST,
SO_COST_DIFFERENCE,
NULL,
EXPENSE_AMOUNT,
NULL,
NULL,
NULL,
LOT_NUMBER,
SO_HEADER_ID,
NULL,
NULL,
NULL,
NULL,
NULL,
NULL,
NULL,
NULL,
NULL,
NULL,
NULL,
NULL,
NULL,
NULL,
SEGMENT1,
SEGMENT2,
NULL,
NULL,
SEGMENT5,
SEGMENT6,
SEGMENT7,
NULL,
ITEM_TYPE,
CREATION_DATE,
CREATED_BY,
LAST_UPDATED_BY,
LAST_UPDATE_DATE,
LAST_UPDATE_LOGIN,
NULL,
ATTRIBUTE1,
NULL,
NULL,
NULL,
NULL,
NULL,
NULL,
NULL,
NULL,
NULL,
NULL,
NULL,
NULL,
NULL,
ATTRIBUTE15,
NULL,
NULL,
NULL,
NULL,
NULL,
NULL,
NULL,
SO_NUMBER,
SO_LINE_NUM,
NULL,
NULL,
NULL,
NULL,
INV_ITEM_DESCRIPTION,
NULL,
NULL,
NULL,
NULL,
NULL,
NULL,
NULL,
INV_ITEM_NUMBER,
RA_TRX_TYPE_NAME,
NULL
FROM (SELECT /*+ leading(hou) no_merge(hou) full(nh) */
(SELECT HOU.ORGANIZATION_ID
FROM HOU, FSP
WHERE HOU.ORGANIZATION_ID = FSP.ORG_ID
AND FSP.INVENTORY_ORGANIZATION_ID = NH.INV_ORG_ID
AND ROWNUM = 1) ORG_ID,
GCC.SEGMENT2 ATTRIBUTE15,
GCC.SEGMENT6 ITEM_CATEGORY,
(SELECT T.DESCRIPTION
FROM T, B
WHERE B.FLEX_VALUE_ID = T.FLEX_VALUE_ID
AND T.LANGUAGE = 'US'
AND B.FLEX_VALUE_SET_ID = 1014879
AND B.FLEX_VALUE = GCC.SEGMENT6) ITEM_CATEGORY_DESC,
GCC.SEGMENT2 DEPARTMENT_CODE,
(SELECT FFV.DESCRIPTION
FROM FFV
WHERE FFV.FLEX_VALUE_SET_ID = 1014875
AND FFV.FLEX_VALUE = GCC.SEGMENT2) DEPARTMENT_NAME,
GCC.SEGMENT7 PROJECT_NUMBER,
SUM(NVL(NH.UPDATE_AMOUNT, 0)) SO_COST,
DECODE(GCC.SEGMENT5, '0', NULL, SUM(NVL(NH.UPDATE_AMOUNT, 0))) INTERNAL_SO_COST,
GCC.SEGMENT1,
GCC.SEGMENT2,
GCC.SEGMENT5,
GCC.SEGMENT6,
GCC.SEGMENT7,
SYSDATE CREATION_DATE,
FND_GLOBAL.USER_ID CREATED_BY,
FND_GLOBAL.USER_ID LAST_UPDATED_BY,
SYSDATE LAST_UPDATE_DATE,
FND_GLOBAL.LOGIN_ID LAST_UPDATE_LOGIN,
0 EXPENSE_AMOUNT,
OOH.ORDER_NUMBER SO_NUMBER,
-1 * SUM(NVL(NH.UPDATE_AMOUNT, 0)) SO_COST_DIFFERENCE,
'成本更新' LOT_NUMBER,
(SELECT NVL(B.ATTRIBUTE4, B.ATTRIBUTE2)
FROM T, B, HCA
WHERE B.FLEX_VALUE_ID = T.FLEX_VALUE_ID
AND T.LANGUAGE = 'US'
AND B.FLEX_VALUE_SET_ID = 1014957
AND B.FLEX_VALUE = HCA.ATTRIBUTE11
AND OOL.SOLD_TO_ORG_ID = HCA.CUST_ACCOUNT_ID
AND ROWNUM = 1) SO_AREA,
(SELECT JRSE.RESOURCE_NAME
FROM JRS, JRSE
WHERE JRS.SALESREP_ID = OOH.SALESREP_ID
AND JRS.RESOURCE_ID = JRSE.RESOURCE_ID
AND JRSE.LANGUAGE = 'US'
AND ROWNUM = 1) SALES_PERSON,
NULL SO_TYPE,
(SELECT OTT.DESCRIPTION
FROM OTT
WHERE OOH.ORDER_TYPE_ID = OTT.TRANSACTION_TYPE_ID
AND OTT.LANGUAGE = USERENV('LANG')
AND ROWNUM = 1) RA_TRX_TYPE_NAME,
(SELECT MSIB.DESCRIPTION
FROM MSIB
WHERE OOL.INVENTORY_ITEM_ID = MSIB.INVENTORY_ITEM_ID
AND OOL.SHIP_FROM_ORG_ID = MSIB.ORGANIZATION_ID
AND ROWNUM = 1) INV_ITEM_DESCRIPTION,
OOL.ORDERED_ITEM INV_ITEM_NUMBER,
(SELECT MDEV.ELEMENT_VALUE
FROM MSIB, MDEV
WHERE OOL.INVENTORY_ITEM_ID = MSIB.INVENTORY_ITEM_ID
AND OOL.SHIP_FROM_ORG_ID = MSIB.ORGANIZATION_ID
AND MDEV.ELEMENT_NAME(+) = '商品简称'
AND MDEV.INVENTORY_ITEM_ID(+) = MSIB.INVENTORY_ITEM_ID
AND ROWNUM = 1) ITEM_TYPE,
(SELECT PAF.PER_NAME
FROM PAF, JR
WHERE JR.PERSON_ID = PAF.PERSON_ID
AND NVL(OOH.ATTRIBUTE3, OOH.SALESREP_ID) = JR.SALESREP_ID(+)
AND OOH.ORG_ID = JR.ORG_ID(+)
AND ROWNUM = 1) ATTRIBUTE1,
OOL.LINE_NUMBER || '.' || OOL.SHIPMENT_NUMBER SO_LINE_NUM,
OOH.HEADER_ID SO_HEADER_ID,
null SESSION_ID
FROM NH, GCC, OOL, OOH, FLV, HOU
WHERE NH.UPDATE_CCID = GCC.CODE_COMBINATION_ID
AND OOL.LINE_ID = NH.SO_LINE_ID
AND NH.SO_HEADER_ID = OOL.HEADER_ID
AND OOL.HEADER_ID = OOH.HEADER_ID
AND GCC.SEGMENT2 > '0'
AND GCC.SEGMENT3 >= FLV.MEANING
AND GCC.SEGMENT3 <= FLV.DESCRIPTION
AND DECODE(null, NULL, '-1', GCC.SEGMENT7) =
DECODE(null, NULL, '-1', null)
AND FLV.LOOKUP_TYPE = 'XY_AR_COST_SUBJECT'
AND NVL(FLV.ENABLED_FLAG, 'N') = 'Y'
AND SYSDATE BETWEEN NVL(FLV.START_DATE_ACTIVE, SYSDATE - 1) AND
NVL(FLV.END_DATE_ACTIVE, SYSDATE + 1)
AND GCC.SUMMARY_FLAG = 'N'
AND NH.GL_IF_FLAG = 'OK'
AND HOU.SHORT_CODE = GCC.SEGMENT1
AND HOU.SET_OF_BOOKS_ID = 2022
AND (GCC.SEGMENT1 = null OR null IS NULL)
AND NH.ATTRIBUTE2 >= '2022-01'
AND NH.ATTRIBUTE2 <= '2022-01'
AND (NH.INV_ORG_ID = null OR null IS NULL)
AND OOH.ORG_ID = HOU.ORGANIZATION_ID
AND OOL.ORG_ID = HOU.ORGANIZATION_ID
AND XY_COM_DEPT_READ.COM_READ(OOH.ORG_ID) = 'Y'
AND (OOH.ORG_ID = null OR null IS NULL)
AND (OOL.ORG_ID = null OR null IS NULL)
GROUP BY NH.INV_ORG_ID,
GCC.SEGMENT2,
GCC.SEGMENT6,
GCC.SEGMENT7,
GCC.SEGMENT1,
GCC.SEGMENT5,
OOH.ORDER_NUMBER,
OOL.SOLD_TO_ORG_ID,
OOH.SALESREP_ID,
OOH.ORDER_TYPE_ID,
OOL.INVENTORY_ITEM_ID,
OOL.SHIP_FROM_ORG_ID,
OOL.ORDERED_ITEM,
NVL(OOH.ATTRIBUTE3, OOH.SALESREP_ID),
OOH.ORG_ID,
OOL.LINE_NUMBER || '.' || OOL.SHIPMENT_NUMBER,
OOH.HEADER_ID)执行计划比较长就不贴了，这是一条INSERT INTO…SELECT…GROUP BY语句
一共要插入54316行数据，GROUP BY之前有2685697条数据
SQL> SELECT /*+ leading(hou) no_merge(hou) full(nh) */
2 count(*)
3 FROM NH,
4 GCC,
5 OOL,
6 OOH,
7 FLV,
8 HOU
9 WHERE NH.UPDATE_CCID = GCC.CODE_COMBINATION_ID
10 AND OOL.LINE_ID = NH.SO_LINE_ID
11 AND NH.SO_HEADER_ID = OOL.HEADER_ID
12 AND OOL.HEADER_ID = OOH.HEADER_ID
13 AND GCC.SEGMENT2 > '0'
14 AND GCC.SEGMENT3 >= FLV.MEANING
15 AND GCC.SEGMENT3 <= FLV.DESCRIPTION
16 AND DECODE(null, NULL, '-1', GCC.SEGMENT7) =
17 DECODE(null, NULL, '-1', null)
18 AND FLV.LOOKUP_TYPE = 'XY_AR_COST_SUBJECT'
19 AND NVL(FLV.ENABLED_FLAG, 'N') = 'Y'
20 AND SYSDATE BETWEEN
21 NVL(FLV.START_DATE_ACTIVE, SYSDATE - 1) AND
22 NVL(FLV.END_DATE_ACTIVE, SYSDATE + 1)
23 AND GCC.SUMMARY_FLAG = 'N'
24 AND NH.GL_IF_FLAG = 'OK'
25 AND HOU.SHORT_CODE = GCC.SEGMENT1
26 AND HOU.SET_OF_BOOKS_ID = 2022
27 AND (GCC.SEGMENT1 = null OR null IS NULL)
28 AND NH.ATTRIBUTE2 >= '2022-01'
29 AND NH.ATTRIBUTE2 <= '2022-01'
30 AND (NH.INV_ORG_ID = null OR null IS NULL)
31 AND OOH.ORG_ID = HOU.ORGANIZATION_ID
32 AND OOL.ORG_ID = HOU.ORGANIZATION_ID
33 AND XY_COM_DEPT_READ.COM_READ(OOH.ORG_ID) = 'Y'
34 AND (OOH.ORG_ID = null OR null IS NULL)
35 AND (OOL.ORG_ID = null OR null IS NULL)
36 ;
SQL>
COUNT(*)
----------
2685697
Executed in 40.204 secondsSQL语句中的where条件部分有个自定义函数，这个自定义函数是用来做权限校验的
XY_COM_DEPT_READ.COM_READ(OOH.ORG_ID) = 'Y'
去掉自定义函数看一下速度
SQL> SELECT /*+ leading(hou) no_merge(hou) full(nh) */
2 count(*)
3 FROM NH,
4 GCC,
5 OOL,
6 OOH,
7 FLV,
8 HOU
9 WHERE NH.UPDATE_CCID = GCC.CODE_COMBINATION_ID
10 AND OOL.LINE_ID = NH.SO_LINE_ID
11 AND NH.SO_HEADER_ID = OOL.HEADER_ID
12 AND OOL.HEADER_ID = OOH.HEADER_ID
13 AND GCC.SEGMENT2 > '0'
14 AND GCC.SEGMENT3 >= FLV.MEANING
15 AND GCC.SEGMENT3 <= FLV.DESCRIPTION
16 AND DECODE(null, NULL, '-1', GCC.SEGMENT7) =
17 DECODE(null, NULL, '-1', null)
18 AND FLV.LOOKUP_TYPE = 'XY_AR_COST_SUBJECT'
19 AND NVL(FLV.ENABLED_FLAG, 'N') = 'Y'
20 AND SYSDATE BETWEEN
21 NVL(FLV.START_DATE_ACTIVE, SYSDATE - 1) AND
22 NVL(FLV.END_DATE_ACTIVE, SYSDATE + 1)
23 AND GCC.SUMMARY_FLAG = 'N'
24 AND NH.GL_IF_FLAG = 'OK'
25 AND HOU.SHORT_CODE = GCC.SEGMENT1
26 AND HOU.SET_OF_BOOKS_ID = 2022
27 AND (GCC.SEGMENT1 = null OR null IS NULL)
28 AND NH.ATTRIBUTE2 >= '2022-01'
29 AND NH.ATTRIBUTE2 <= '2022-01'
30 AND (NH.INV_ORG_ID = null OR null IS NULL)
31 AND OOH.ORG_ID = HOU.ORGANIZATION_ID
32 AND OOL.ORG_ID = HOU.ORGANIZATION_ID
33 --AND XY_COM_DEPT_READ.COM_READ(OOH.ORG_ID) = 'Y'
34 AND (OOH.ORG_ID = null OR null IS NULL)
35 AND (OOL.ORG_ID = null OR null IS NULL)
36 ;
COUNT(*)
----------
2685697
Executed in 10.853 seconds不加自定义函数10秒跑完，加了40秒
因为原始SQL有GROUP BY，而且GROUP BY的列有OOH.ORG_ID，所以可以将
XY_COM_DEPT_READ.COM_READ(OOH.ORG_ID) = 'Y’放到GROUP BY之后再过滤
于是原始SQL可以改写为：
INSERT INTO TMP
SELECT 1,
ORG_ID,
SESSION_ID,
NULL,
NULL,
NULL,
DEPARTMENT_CODE,
NULL,
DEPARTMENT_NAME,
NULL,
SALES_PERSON,
NULL,
PROJECT_NUMBER,
NULL,
NULL,
NULL,
NULL,
NULL,
NULL,
NULL,
ITEM_CATEGORY,
ITEM_CATEGORY_DESC,
NULL,
SO_TYPE,
SO_AREA,
NULL,
NULL,
NULL,
NULL,
NULL,
NULL,
SO_COST,
INTERNAL_SO_COST,
SO_COST_DIFFERENCE,
NULL,
EXPENSE_AMOUNT,
NULL,
NULL,
NULL,
LOT_NUMBER,
SO_HEADER_ID,
NULL,
NULL,
NULL,
NULL,
NULL,
NULL,
NULL,
NULL,
NULL,
NULL,
NULL,
NULL,
NULL,
NULL,
SEGMENT1,
SEGMENT2,
NULL,
NULL,
SEGMENT5,
SEGMENT6,
SEGMENT7,
NULL,
ITEM_TYPE,
CREATION_DATE,
CREATED_BY,
LAST_UPDATED_BY,
LAST_UPDATE_DATE,
LAST_UPDATE_LOGIN,
NULL,
ATTRIBUTE1,
NULL,
NULL,
NULL,
NULL,
NULL,
NULL,
NULL,
NULL,
NULL,
NULL,
NULL,
NULL,
NULL,
ATTRIBUTE15,
NULL,
NULL,
NULL,
NULL,
NULL,
NULL,
NULL,
SO_NUMBER,
SO_LINE_NUM,
NULL,
NULL,
NULL,
NULL,
INV_ITEM_DESCRIPTION,
NULL,
NULL,
NULL,
NULL,
NULL,
NULL,
NULL,
INV_ITEM_NUMBER,
RA_TRX_TYPE_NAME,
NULL
FROM (SELECT /*+ leading(hou) no_merge(hou) full(nh) */
(SELECT HOU.ORGANIZATION_ID
FROM HOU, FSP
WHERE HOU.ORGANIZATION_ID = FSP.ORG_ID
AND FSP.INVENTORY_ORGANIZATION_ID = NH.INV_ORG_ID
AND ROWNUM = 1) ORG_ID,
GCC.SEGMENT2 ATTRIBUTE15,
GCC.SEGMENT6 ITEM_CATEGORY,
(SELECT T.DESCRIPTION
FROM T, B
WHERE B.FLEX_VALUE_ID = T.FLEX_VALUE_ID
AND T.LANGUAGE = 'US'
AND B.FLEX_VALUE_SET_ID = 1014879
AND B.FLEX_VALUE = GCC.SEGMENT6) ITEM_CATEGORY_DESC,
GCC.SEGMENT2 DEPARTMENT_CODE,
(SELECT FFV.DESCRIPTION
FROM FFV
WHERE FFV.FLEX_VALUE_SET_ID = 1014875
AND FFV.FLEX_VALUE = GCC.SEGMENT2) DEPARTMENT_NAME,
GCC.SEGMENT7 PROJECT_NUMBER,
SUM(NVL(NH.UPDATE_AMOUNT, 0)) SO_COST,
DECODE(GCC.SEGMENT5, '0', NULL, SUM(NVL(NH.UPDATE_AMOUNT, 0))) INTERNAL_SO_COST,
GCC.SEGMENT1,
GCC.SEGMENT2,
GCC.SEGMENT5,
GCC.SEGMENT6,
GCC.SEGMENT7,
SYSDATE CREATION_DATE,
FND_GLOBAL.USER_ID CREATED_BY,
FND_GLOBAL.USER_ID LAST_UPDATED_BY,
SYSDATE LAST_UPDATE_DATE,
FND_GLOBAL.LOGIN_ID LAST_UPDATE_LOGIN,
0 EXPENSE_AMOUNT,
OOH.ORDER_NUMBER SO_NUMBER,
-1 * SUM(NVL(NH.UPDATE_AMOUNT, 0)) SO_COST_DIFFERENCE,
'成本更新' LOT_NUMBER,
(SELECT NVL(B.ATTRIBUTE4, B.ATTRIBUTE2)
FROM T, B, HCA
WHERE B.FLEX_VALUE_ID = T.FLEX_VALUE_ID
AND T.LANGUAGE = 'US'
AND B.FLEX_VALUE_SET_ID = 1014957
AND B.FLEX_VALUE = HCA.ATTRIBUTE11
AND OOL.SOLD_TO_ORG_ID = HCA.CUST_ACCOUNT_ID
AND ROWNUM = 1) SO_AREA,
(SELECT JRSE.RESOURCE_NAME
FROM JRS, JRSE
WHERE JRS.SALESREP_ID = OOH.SALESREP_ID
AND JRS.RESOURCE_ID = JRSE.RESOURCE_ID
AND JRSE.LANGUAGE = 'US'
AND ROWNUM = 1) SALES_PERSON,
NULL SO_TYPE,
(SELECT OTT.DESCRIPTION
FROM OTT
WHERE OOH.ORDER_TYPE_ID = OTT.TRANSACTION_TYPE_ID
AND OTT.LANGUAGE = USERENV('LANG')
AND ROWNUM = 1) RA_TRX_TYPE_NAME,
(SELECT MSIB.DESCRIPTION
FROM MSIB
WHERE OOL.INVENTORY_ITEM_ID = MSIB.INVENTORY_ITEM_ID
AND OOL.SHIP_FROM_ORG_ID = MSIB.ORGANIZATION_ID
AND ROWNUM = 1) INV_ITEM_DESCRIPTION,
OOL.ORDERED_ITEM INV_ITEM_NUMBER,
(SELECT MDEV.ELEMENT_VALUE
FROM MSIB, MDEV
WHERE OOL.INVENTORY_ITEM_ID = MSIB.INVENTORY_ITEM_ID
AND OOL.SHIP_FROM_ORG_ID = MSIB.ORGANIZATION_ID
AND MDEV.ELEMENT_NAME(+) = '商品简称'
AND MDEV.INVENTORY_ITEM_ID(+) = MSIB.INVENTORY_ITEM_ID
AND ROWNUM = 1) ITEM_TYPE,
(SELECT PAF.PER_NAME
FROM PAF, JR
WHERE JR.PERSON_ID = PAF.PERSON_ID
AND NVL(OOH.ATTRIBUTE3, OOH.SALESREP_ID) = JR.SALESREP_ID(+)
AND OOH.ORG_ID = JR.ORG_ID(+)
AND ROWNUM = 1) ATTRIBUTE1,
OOL.LINE_NUMBER || '.' || OOL.SHIPMENT_NUMBER SO_LINE_NUM,
OOH.HEADER_ID SO_HEADER_ID,
null SESSION_ID
FROM NH, GCC, OOL, OOH, FLV, HOU
WHERE NH.UPDATE_CCID = GCC.CODE_COMBINATION_ID
AND OOL.LINE_ID = NH.SO_LINE_ID
AND NH.SO_HEADER_ID = OOL.HEADER_ID
AND OOL.HEADER_ID = OOH.HEADER_ID
AND GCC.SEGMENT2 > '0'
AND GCC.SEGMENT3 >= FLV.MEANING
AND GCC.SEGMENT3 <= FLV.DESCRIPTION
AND DECODE(null, NULL, '-1', GCC.SEGMENT7) =
DECODE(null, NULL, '-1', null)
AND FLV.LOOKUP_TYPE = 'XY_AR_COST_SUBJECT'
AND NVL(FLV.ENABLED_FLAG, 'N') = 'Y'
AND SYSDATE BETWEEN NVL(FLV.START_DATE_ACTIVE, SYSDATE - 1) AND
NVL(FLV.END_DATE_ACTIVE, SYSDATE + 1)
AND GCC.SUMMARY_FLAG = 'N'
AND NH.GL_IF_FLAG = 'OK'
AND HOU.SHORT_CODE = GCC.SEGMENT1
AND HOU.SET_OF_BOOKS_ID = 2022
AND (GCC.SEGMENT1 = null OR null IS NULL)
AND NH.ATTRIBUTE2 >= '2022-01'
AND NH.ATTRIBUTE2 <= '2022-01'
AND (NH.INV_ORG_ID = null OR null IS NULL)
AND OOH.ORG_ID = HOU.ORGANIZATION_ID
AND OOL.ORG_ID = HOU.ORGANIZATION_ID
--AND XY_COM_DEPT_READ.COM_READ(OOH.ORG_ID) = 'Y'
AND (OOH.ORG_ID = null OR null IS NULL)
AND (OOL.ORG_ID = null OR null IS NULL)
GROUP BY NH.INV_ORG_ID,
GCC.SEGMENT2,
GCC.SEGMENT6,
GCC.SEGMENT7,
GCC.SEGMENT1,
GCC.SEGMENT5,
OOH.ORDER_NUMBER,
OOL.SOLD_TO_ORG_ID,
OOH.SALESREP_ID,
OOH.ORDER_TYPE_ID,
OOL.INVENTORY_ITEM_ID,
OOL.SHIP_FROM_ORG_ID,
OOL.ORDERED_ITEM,
NVL(OOH.ATTRIBUTE3, OOH.SALESREP_ID),
OOH.ORG_ID,
OOL.LINE_NUMBER || '.' || OOL.SHIPMENT_NUMBER,
OOH.HEADER_ID)
WHERE XY_COM_DEPT_READ.COM_READ(ORG_ID) = 'Y'
54316 rows inserted
Executed in 17.684 seconds改写之后从47秒降低到17秒，有的读者可能会有疑问，这个SQL恰好是GROUP BY的列有OOH.ORG_ID
如果GROUP BY的列中没有有OOH.ORG_ID，怎么改写呢？其实也很简单，可以这样改写
SQL> SELECT /*+ leading(hou) no_merge(hou) full(nh) */
2 count(*)
3 FROM NH,
4 GCC,
5 OOL,
6 OOH,
7 FLV,
8 HOU
9 WHERE NH.UPDATE_CCID = GCC.CODE_COMBINATION_ID
10 AND OOL.LINE_ID = NH.SO_LINE_ID
11 AND NH.SO_HEADER_ID = OOL.HEADER_ID
12 AND OOL.HEADER_ID = OOH.HEADER_ID
13 AND GCC.SEGMENT2 > '0'
14 AND GCC.SEGMENT3 >= FLV.MEANING
15 AND GCC.SEGMENT3 <= FLV.DESCRIPTION
16 AND DECODE(null, NULL, '-1', GCC.SEGMENT7) =
17 DECODE(null, NULL, '-1', null)
18 AND FLV.LOOKUP_TYPE = 'XY_AR_COST_SUBJECT'
19 AND NVL(FLV.ENABLED_FLAG, 'N') = 'Y'
20 AND SYSDATE BETWEEN
21 NVL(FLV.START_DATE_ACTIVE, SYSDATE - 1) AND
22 NVL(FLV.END_DATE_ACTIVE, SYSDATE + 1)
23 AND GCC.SUMMARY_FLAG = 'N'
24 AND NH.GL_IF_FLAG = 'OK'
25 AND HOU.SHORT_CODE = GCC.SEGMENT1
26 AND HOU.SET_OF_BOOKS_ID = 2022
27 AND (GCC.SEGMENT1 = null OR null IS NULL)
28 AND NH.ATTRIBUTE2 >= '2022-01'
29 AND NH.ATTRIBUTE2 <= '2022-01'
30 AND (NH.INV_ORG_ID = null OR null IS NULL)
31 AND OOH.ORG_ID = HOU.ORGANIZATION_ID
32 AND OOL.ORG_ID = HOU.ORGANIZATION_ID
33 --AND XY_COM_DEPT_READ.COM_READ(OOH.ORG_ID) = 'Y'
34 AND EXISTS (SELECT /*+ NO_UNNEST */ NULL FROM DUAL WHERE XY_COM_DEPT_READ.COM_READ(OOH.ORG_ID) = 'Y')
35 AND (OOH.ORG_ID = null OR null IS NULL)
36 AND (OOL.ORG_ID = null OR null IS NULL)
37 ;
COUNT(*)
----------
2685697
Executed in 15.281 seconds改写为EXISTS(SELECT /*+ NO_UNNEST */)让其走FILTER，因为FILTER有去重功能，所以走FILTER也会减少函数被调用的次数
总结：最优的方法应该是将自定义函数中的SQL提取出来，改写为表连接，但是有时候自定义函数特别复杂，改写为表连接相比减少函数被调用次数性能提升也没有特别大的提升，所以就不必要去折腾
以上是关于where条件中有权限校验的自定义函数优化方法的主要内容，如果未能解决你的问题，请参考以下文章