[20190918]关于函数索引问题.txt
Posted lfree
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[20190918]关于函数索引问题.txt相关的知识,希望对你有一定的参考价值。
[20190918]关于函数索引问题.txt
1.环境:
SCOTT@test01p> @ ver1
PORT_STRING VERSION BANNER CON_ID
------------------------------ -------------- -------------------------------------------------------------------------------- ----------
IBMPC/WIN_NT64-9.1.0 12.2.0.1.0 Oracle Database 12c Enterprise Edition Release 12.2.0.1.0 - 64bit Production 0
create table t as select 1 id1, rownum id2 ,‘test‘ name from dual connect by level<1e4;
insert into t values (1e4,1e4,‘abcd‘);
commit ;
create index if_t_id2 on t(decode(id1, 1, to_number(null), id2));
--//分析表略。
--//简单说明,使用to_number(null)保证返回数据类型是number类型的NULL值。
2.测试:
SCOTT@test01p> select column_name,data_type from user_tab_cols where table_name = ‘T‘ ;
COLUMN_NAME DATA_TYPE
-------------------- ----------
ID1 NUMBER
ID2 NUMBER
NAME CHAR
SYS_NC00004$ NUMBER
--//增加一个隐含字段SYS_NC00004$.返回数据类型是number类型.
SCOTT@test01p> alter session set statistics_level = all;
Session altered.
SCOTT@test01p> select * from t where decode(id1, 1,to_number(null), id2) = 1e4;
ID1 ID2 NAME
---------- ---------- --------------------
10000 10000 abcd
SCOTT@test01p> @ dpc ‘‘ ‘‘
PLAN_TABLE_OUTPUT
-------------------------------------
SQL_ID 7srfk2yjdxx49, child number 0
-------------------------------------
select * from t where decode(id1, 1,to_number(null), id2) = 1e4
Plan hash value: 1601196873
--------------------------------------------------------------------------------------------------------------------
| Id | Operation | Name | Starts | E-Rows |E-Bytes| Cost (%CPU)| E-Time | A-Rows | A-Time | Buffers |
--------------------------------------------------------------------------------------------------------------------
| 0 | SELECT STATEMENT | | 1 | | | 10 (100)| | 1 |00:00:00.01 | 31 |
|* 1 | TABLE ACCESS FULL| T | 1 | 100 | 1200 | 10 (0)| 00:00:01 | 1 |00:00:00.01 | 31 |
--------------------------------------------------------------------------------------------------------------------
Query Block Name / Object Alias (identified by operation id):
-------------------------------------------------------------
1 - SEL$1 / T@SEL$1
Predicate Information (identified by operation id):
---------------------------------------------------
1 - filter(DECODE("ID1",1,NULL,"ID2")=10000)
--//你可以发现实际上filter(DECODE("ID1",1,NULL,"ID2")=10000).
SCOTT@test01p> select * from user_ind_expressions where index_name = ‘IF_T_ID2‘
2 @prxx
==============================
INDEX_NAME : IF_T_ID2
TABLE_NAME : T
COLUMN_EXPRESSION : DECODE("ID1",1,NULL,"ID2")
COLUMN_POSITION : 1
PL/SQL procedure successfully completed.
--//你可以发现我建立的函数索引的表达式与保存的不一致.
--//尝试改写看看呢?
SCOTT@test01p> select * from t where decode(id1, 1,null, id2) = 1e4;
ID1 ID2 NAME
---------- ---------- --------------------
10000 10000 abcd
SCOTT@test01p> @ dpc ‘‘ ‘‘
PLAN_TABLE_OUTPUT
-------------------------------------
SQL_ID by0609fp41hy2, child number 0
-------------------------------------
select * from t where decode(id1, 1,null, id2) = 1e4
Plan hash value: 1130968923
------------------------------------------------------------------------------------------------------------------------------------------
| Id | Operation | Name | Starts | E-Rows |E-Bytes| Cost (%CPU)| E-Time | A-Rows | A-Time | Buffers |
------------------------------------------------------------------------------------------------------------------------------------------
| 0 | SELECT STATEMENT | | 1 | | | 2 (100)| | 1 |00:00:00.01 | 2 |
| 1 | TABLE ACCESS BY INDEX ROWID BATCHED| T | 1 | 1 | 13 | 2 (0)| 00:00:01 | 1 |00:00:00.01 | 2 |
|* 2 | INDEX RANGE SCAN | IF_T_ID2 | 1 | 1 | | 1 (0)| 00:00:01 | 1 |00:00:00.01 | 1 |
------------------------------------------------------------------------------------------------------------------------------------------
Query Block Name / Object Alias (identified by operation id):
-------------------------------------------------------------
1 - SEL$1 / T@SEL$1
2 - SEL$1 / T@SEL$1
Predicate Information (identified by operation id):
---------------------------------------------------
2 - access("T"."SYS_NC00004$"=10000)
--//这样可以使用索引,而这样明显存在1个错误,按照Yangtingkun以前blog介绍,这样的返回类型是字符型.因为NULL没有明确指定
--//缺省类型是varchar2类型.而实际现在是number类型.是因为SYS_NC00004$是NUMBER类型.
3.继续测试:
--//如果rebuild online索引呢?
SCOTT@test01p> alter index IF_T_ID2 rebuild online ;
Index altered.
SCOTT@test01p> select * from t where decode(id1, 1,null, id2) = 1e4;
ID1 ID2 NAME
---------- ---------- --------------------
10000 10000 abcd
SCOTT@test01p> @ dpc ‘‘ ‘‘
PLAN_TABLE_OUTPUT
-------------------------------------
SQL_ID by0609fp41hy2, child number 0
-------------------------------------
select * from t where decode(id1, 1,null, id2) = 1e4
Plan hash value: 1601196873
--------------------------------------------------------------------------------------------------------------------
| Id | Operation | Name | Starts | E-Rows |E-Bytes| Cost (%CPU)| E-Time | A-Rows | A-Time | Buffers |
--------------------------------------------------------------------------------------------------------------------
| 0 | SELECT STATEMENT | | 1 | | | 11 (100)| | 1 |00:00:00.01 | 31 |
|* 1 | TABLE ACCESS FULL| T | 1 | 1 | 13 | 11 (10)| 00:00:01 | 1 |00:00:00.01 | 31 |
--------------------------------------------------------------------------------------------------------------------
Query Block Name / Object Alias (identified by operation id):
-------------------------------------------------------------
1 - SEL$1 / T@SEL$1
Predicate Information (identified by operation id):
---------------------------------------------------
1 - filter(TO_NUMBER(DECODE("ID1",1,NULL,TO_CHAR("ID2")))=10000)
--//注意看过滤条件,发生了隐式转换.前面加上了TO_NUMBER.
SCOTT@test01p> select column_name,data_type from user_tab_cols where table_name = ‘T‘ ;
COLUMN_NAME DATA_TYPE
-------------------- --------------------
ID1 NUMBER
ID2 NUMBER
NAME CHAR
SYS_NC00004$ VARCHAR2
--//重建索引后,隐含字段SYS_NC00004$的数据类型对比前面的情况发生了变化,变为varchar2类型.
--//要保证使用索引应该写成如下:
SCOTT@test01p> select * from t where decode(id1, 1,null, id2) = to_char(1e4);
ID1 ID2 NAME
---------- ---------- --------------------
10000 10000 abcd
SCOTT@test01p> @ dpc ‘‘ ‘‘
PLAN_TABLE_OUTPUT
-------------------------------------
SQL_ID 63zyt23ufr2xa, child number 0
-------------------------------------
select * from t where decode(id1, 1,null, id2) = to_char(1e4)
Plan hash value: 1130968923
------------------------------------------------------------------------------------------------------------------------------------------
| Id | Operation | Name | Starts | E-Rows |E-Bytes| Cost (%CPU)| E-Time | A-Rows | A-Time | Buffers |
------------------------------------------------------------------------------------------------------------------------------------------
| 0 | SELECT STATEMENT | | 1 | | | 2 (100)| | 1 |00:00:00.01 | 2 |
| 1 | TABLE ACCESS BY INDEX ROWID BATCHED| T | 1 | 1 | 13 | 2 (0)| 00:00:01 | 1 |00:00:00.01 | 2 |
|* 2 | INDEX RANGE SCAN | IF_T_ID2 | 1 | 1 | | 1 (0)| 00:00:01 | 1 |00:00:00.01 | 1 |
------------------------------------------------------------------------------------------------------------------------------------------
Query Block Name / Object Alias (identified by operation id):
-------------------------------------------------------------
1 - SEL$1 / T@SEL$1
2 - SEL$1 / T@SEL$1
Predicate Information (identified by operation id):
---------------------------------------------------
2 - access("T"."SYS_NC00004$"=‘10000‘)
以上是关于[20190918]关于函数索引问题.txt的主要内容,如果未能解决你的问题,请参考以下文章
pandas读取csv数据index_col参数指定作为行索引的数据列索引列表形成复合(多层)行索引使用reset_index函数把行索引重置为列数据(原来的行索名称转化为列索引的最外层)
pandas读取csv数据index_col参数指定作为行索引的数据列索引列表形成复合(多层)行索引使用reset_index函数把行索引重置为列数据(原来的行索名称转化为列索引的最外层)
pandas读取csv数据使用reset_index函数把行索引重置为列数据(col_level参数设置原来的行索名称转化为列索引的指定层索引col_fill参数为缺失内容的层进行填充)