一、抽象数据类型
1、创建类型
–地址类型
CREATE OR REPLACE TYPE AddressType AS OBJECT
(
Country varchar2(15),
City varchar2(20),
Street varchar2(30)
);
2、类型嵌套
–创建基于前一个类型的新的抽象数据类型:巨星类型
CREATE OR REPLACE TYPE SuperStarType AS OBJECT
(
StarName varchar2(30),
Address AddressType
);
3、基于抽象类型创建关系表
CREATE TABLE SuperStar
(
StarID varchar(10),
Star SuperStarType
);
4、基于抽象类型创建对象表
CREATE TABLE SuperStarObj of SuperStarType;
5、使用构造方法在表中插入记录
INSERT INTO SuperStar VALUES('001',SuperStarType('Zidane',AddressType('France','Paris','People Street NO.1')));
6、查询表中记录
(1)SQL> SELECT * FROM SuperStar;
STARID
———-
STAR(STARNAME, ADDRESS(COUNTRY, CITY, STREET))
——————————————————————————–
001
SUPERSTARTYPE('Zidane', ADDRESSTYPE('France', 'Paris', 'People Street NO.1'))
(2)
SELECT s.StarID,s.Star.StarName,s.Star.Address.Country,s.Star.Address.City,s.Star.Address.Street FROM SuperStar s
STARID STAR.STARNAME STAR.ADDRESS.CO STAR.ADDRESS.CITY STAR.ADDRESS.STREET
———- —————————— ————— ——————– ———————
001 Zidane France Paris People Street NO.1
7、抽象数据类型的继承
(1)创建一个类型
CREATE OR REPLACE TYPE PersonType AS OBJECT
(
PersonName varchar(10),
PersonSex varchar(2),
PersonBirth date
) not final;
(2)派生一个类型
CREATE OR REPLACE TYPE StudentType UNDER PersonType
(
StudentNO int,
StudentScore int
);
(3)查看数据字典
SQL> DESC StudentType
StudentType extends SYS.PERSONTYPE
Name
——————————————————————————
PERSONNAME
PERSONSEX
PERSONBIRTH
STUDENTNO
STUDENTSCORE
(4)创建对象表
CREATE TABLE student OF StudentType;
(5)向对象表中插入数据
INSERT INTO student VALUES('Rose','nv',to_date('1983-05-02','yyyy-mm-dd'),1001,98);
(6) 查询数据
SQL> SELECT * FROM student;
PERSONNAME PE PERSONBIR STUDENTNO STUDENTSCORE
———- — ——— ———- ————
Rose nv 02-MAY-83 1001 98
二、可变数组
1、创建带有可变数组的表
(1)创建可变数组的基类型
CREATE OR REPLACE TYPE MingXiType AS OBJECT
(
GoodID varchar2(20),
InCount int,
ProviderID varchar(20)
);
(2)创建嵌套项类型的可变数组
CREATE OR REPLACE TYPE arrMingXiType AS VARRAY(100) OF MingXiType;
(3)创建一个主表
CREATE TABLE InStockOrder
(
OrderID varchar(15) Not Null Primary Key,
InDate date,
OperatorID varchar(15),
MingXi arrMingXiType
);
2、操作可变数组
(1)插入数据
INSERT INTO InStockOrder
VALUES('200710110001',TO_DATE('2007-10-11','YYYY-MM-DD'),'007',
arrMingXiType(MingXiType('G001',100,'1001'),
MingXiType('G002',888,'1002'))
);
(2)查询数据
SQL> SELECT * FROM InStockOrder;
ORDERID INDATE OPERATORID
————— ——— —————
MINGXI(GOODID, INCOUNT, PROVIDERID)
———————————————————————-
200710110001 11-OCT-07 007
ARRMINGXITYPE(MINGXITYPE('G001', 100, '1001'), MINGXITYPE('G002', 888, '1002')
(3)使用Table()函数
SQL> SELECT * FROM Table(SELECT t.MingXi FROM InStockOrder t
WHERE t.OrderID='200710110001');
GOODID INCOUNT PROVIDERID
——————– ———- ——————–
G001 100 1001
G002 888 1002
(4)修改数据
UPDATE InStockOrder
SET MingXi=arrMingXiType(MingXiType('G001',200,'1001'),
MingXiType('G002',8888,'1002'))
WHERE OrderID='200710110001'
注意:不能更新VARRAY中的单个元素,必须更新整个VARRAY
三、嵌套表
1、创建嵌套表
(1)创建嵌套表的基类型
CREATE OR REPLACE TYPE MingXiType AS OBJECT
(
GoodID varchar2(20),
InCount int,
ProviderID varchar(20)
) not final;
(2)创建嵌套表类型
CREATE OR REPLACE TYPE nestMingXiType AS TABLE OF MingXiType;
(3)创建主表,其中一列是嵌套表类型
CREATE TABLE InStockTable
(
OrderID varchar(15) Not Null Primary Key,
InDate date,
OperatorID varchar(15),
MingXi nestMingXiType
) Nested Table MingXi STORE AS MingXiTable;
2、操作嵌套表
(1)向嵌套表中插入记录
INSERT INTO InStockTable
VALUES('20071012001',TO_DATE('2007-10-12','YYYY-MM-DD'),'007',
nestMingXiType(MingXiType('G003',666,'1001'),
MingXiType('G004',888,'1002'),
MingXiType('G005',8888,'1003'))
);
(2)查询数据
SQL> SELECT * FROM InStockTable;
ORDERID INDATE OPERATORID
————— ——— —————
MINGXI(GOODID, INCOUNT, PROVIDERID)
—————————————————————————————————-
20071012001 12-OCT-07 007
NESTMINGXITYPE(MINGXITYPE('G003', 666, '1001'), MINGXITYPE('G004', 888, '1002'), MINGXITYPE('G005', 8888, '1003')
(3)使用Table()函数
SQL> SELECT * FROM Table(SELECT T.MingXi FROM InStockTable t
WHERE OrderID='20071012001')
GOODID INCOUNT PROVIDERID
——————– ———- ——————–
G003 666 1001
G004 888 1002
G005 8888 1003
(4)更新嵌套表中的数据
UPDATE Table(SELECT t.MingXi FROM InStockTable t WHERE OrderID='20071012001') tt
SET tt.InCount=1666 WHERE tt.GoodID='G003';
(5)删除表中数据
DELETE Table(SELECT t.MingXi FROM InStockTable t WHERE OrderID='20071012001') tt
WHERE tt.GoodID='G003'
四、对象表
1、创建对象表
CREATE TABLE ObjectTable OF MingXiType;
2、向表中插入数据
INSERT INTO ObjectTable VALUES('G001',500,'P005');
INSERT INTO ObjectTable VALUES('G002',1000,'P008');
3、查询对象表中的记录
A 直接查询
SQL> SELECT * FROM ObjectTable;
GOODID INCOUNT PROVIDERID
——————– ———- —————
G001 500 P005
G002 1000 P008
B 用VALUE()函数查询
SQL> SELECT VALUE(O) FROM ObjectTable O;
VALUE(O)(GOODID, INCOUNT, PROVIDERID)
——————————————
MINGXITYPE('G001', 500, 'P005')
MINGXITYPE('G002', 1000, 'P008')
4、查看对象标志符(OID)
A REF操作符引用行对象
SQL> SELECT REF(t) FROM ObjectTable t;
REF(T)
——————————————————————————–
0000280209771F103ED34842478A9C439CDAEFEF6324B0ACF849F14BD7A8B52F4B0297D1C90040A9
5A0000
0000280209A2D3359E0F0C44B3AF652B944F8823F524B0ACF849F14BD7A8B52F4B0297D1C90040A9
5A0001
B 将OID用于创建外键
CREATE TABLE Customer
( CustomerID varchar(10) PRIMARY KEY,
CustomerName varchar(20),
CustomerGoods REF MingXiType SCOPE IS ObjectTable,–引用MingXiType外键,关联的是OID的值
CustomerAddress varchar(20)
);
C 向Customer表中插入数据,此表将从上面创建的对象表中引用数据
INSERT INTO Customer SELECT '007','Yuanhy',REF(O),'France'
FROM ObjectTable O
WHERE GoodID='G001';
D 查询Customer表
SQL> SELECT * FROM Customer;
CUSTOMERID CUSTOMERNAME
———- ——————–
CUSTOMERGOODS
—————————————————————————–
CUSTOMERADDRESS
——————–
007 Yuanhy
0000220208771F103ED34842478A9C439CDAEFEF6324B0ACF849F14BD7A8B52F4B0297D1C9
France
E 用DEREF操作符返回对象的值
SQL> SELECT CustomerID,CustomerName,DEREF(t.CustomerGoods),CustomerAddress
2 FROM Customer t;
CUSTOMERID CUSTOMERNAME
———- ——————–
DEREF(T.CUSTOMERGOODS)(GOODID, INCOUNT, PROVIDERID)
—————————————————————————-
CUSTOMERADDRESS
——————–
007 Yuanhy
MINGXITYPE('G001', 500, 'P005')
France
五、对象视图
将关系表化装成对象表
1、 创建对象视图
A 创建基于关系表父表的对象类型
CREATE OR REPLACE TYPE depttype AS OBJECT
(
deptid number(10),
deptname varchar(30),
loc number(10)
);
B 创建基于关系表的对象视图
CREATE VIEW deptview OF depttype WITH OBJECT OID(deptid) AS
SELECT department_id,department_name,location_id FROM dept;
C 查询视图
SQL> SELECT * FROM deptview;
DEPTID DEPTNAME LOC
———- —————————— ———-
10 Administration 1700
20 Marketing 1800
30 Purchasing 1700
40 Human Resources 2400
50 Shipping 1500
60 IT 1400
70 Public Relations 2700
SQL> select ref(t) from deptview t;
REF(T)
—————————————————————————————————-
00004A038A004667BAC3685B444520A60ED30027E8F25F0000001426010001000100290000000000090604002A00078401FE
00004A038A004667BAC3685B444520A60ED30027E8F25F0000001426010001000100290000000000090604002A00078401FE
00004A038A004667BAC3685B444520A60ED30027E8F25F0000001426010001000100290000000000090604002A00078401FE
00004A038A004667BAC3685B444520A60ED30027E8F25F0000001426010001000100290000000000090604002A00078401FE
00004A038A004667BAC3685B444520A60ED30027E8F25F0000001426010001000100290000000000090604002A00078401FE
00004A038A004667BAC3685B444520A60ED30027E8F25F0000001426010001000100290000000000090604002A00078401FE
00004A038A004667BAC3685B444520A60ED30027E8F25F0000001426010001000100290000000000090604002A00078401FE
2、创建引用视图(类似于关系表创建一个从表)
CREATE VIEW empview AS SELECT MAKE_REF(deptview,department_id) deptOID,employee_id,
first_name,last_name FROM emp;
查询对象视图empview
SQL> SELECT * FROM empview;
DEPTOID
—————————————————————————————————-
EMPLOYEE_ID FIRST_NAME LAST_NAME
———– ——————– ————————-
00004A038A004667BAC3685B444520A60ED30027E8F25F0000001426010001000100290000000000090604002A00078401FE
100 Steven King
00004A038A004667BAC3685B444520A60ED30027E8F25F0000001426010001000100290000000000090604002A00078401FE
101 Neena Kochhar
00004A038A004667BAC3685B444520A60ED30027E8F25F0000001426010001000100290000000000090604002A00078401FE
102 Lex De Haan
00004A038A004667BAC3685B444520A60ED30027E8F25F0000001426010001000100290000000000090604002A00078401FE
103 Alexander Hunold
Oracle对象类型也有属性和方法.
创建对象类型与创建表很相似,只是实际上不为存储的数据分配空间:
不带方法的简单对象类型:
CREATE TYPE type_name as OBJECT (
column_1 type1,
column_2 type2,
…
);
注意:AS OBJECT
创建好对象类型之后,就可以在创建表的时候,使用该类型了,如:
CREATE TYPE HUMAN AS OBJECT(
NAME VARCHAR2(20),
SEX VARCHAR2(1),– F : FEMALE M:MALE
BIRTHDAY DATE,
NOTE VARCHAR2(300)
)
稍后,可以用下面的语句查看:
SELECT * FROM USER_OBJECTS WHERE OBJECT_TYPE = 'TYPE'
CREATE TABLE STUDENTS(
GUID NUMBER NOT NULL,
STUDENTS HUMAN
)
此下省去两个Trigger.
插入数据的时候,可以如下:
INSERT INTO STUDENTS (STUDENT) VALUES (HUMAN('xling','M',TO_DATE('20060101','YYYYMMDD'),'测试'))
注意:HUMAN('xling','M',TO_DATE('20060101','YYYYMMDD'),'测试'),这是个默认的构造函数.
如果想选出性别为女(F)的记录,可以如下:
SELECT * FROM STUDENTS S WHERE S.STUDENT.SEX = 'F'
注意:不能写成:SELECT * FROM STUDENTS WHERE STUDENT.SEX = 'F' 这样会报如下错误:ORA-00904: "STUDENT"."SEX": 标识符无效
对象类型表:每条记录都是对象的表,称为对象类型表.它有两个使用方法:1,用作只有一个对象类型字段的表.2,用作具有对象类型字段的标准关系表.
语法如下:
CREATE TABLE table_name OF object_type;
例如:
CREATE TABLE TMP_STUDENTS OF HUMAN;
用DESC TMP_STUDENTS,可以看到它的字段结构和HUMAN的结构一样.
对象类型表有两个优点:1,从某种程度上简化了对象的使用,因为对象表的字段类型与对象类型是一致的,所以,不需要用对象名来修饰对象属性,可以把数据插入对象类型表,就像插入普通的关系表中一样:
INSERT INTO TMP_STUDENTS VALUES ('xling','M',TO_DATE('20060601','YYYYMMDD'),'对象类型表');
当然也可用如下方法插入:
INSERT INTO TMP_STUDENTS VALUES (HUMAN('snow','F',TO_DATE('20060102','YYYYMMDD'),'用类型的构造函数'));
第二个特点是:对象表是使用对象类型作为模板来创建表的一种便捷方式,它可以确保多个表具有相同的结构.
对象类型表在:USER_TABLES表里是查不到的,而在USER_OBJECTS表里可以查到,而且OBJECT_TYPE = 'TABLE'
类型在定义的时候,会自动包含一个方法,即默认的构造器.构造器的名称与对象的名称相同,它带有变量与对象类型的每个属性相对应.
对象类型的方法:
CREATE TYPE type_name AS OBJECT (
column1 column_type1,
column2 column_type2,
… ,
MEMBER FUNCTION method_name(args_list) RETURN return_type,
…
)
注意:是MEMBER FUNCTION,(当然,也可是MEMBER PROCEDURE,没有返回值)
和包(PACKAGE)一样,如果对象类型有方法的话,还要声明一个BODY:
CREATE TYPE BODY type_name AS
MEMBER FUNCTION method_name RETURN return_type {AS | IS}
variable declareations..
BEGIN
CODE..
RETURN return_value;
END;//END MEMBER FUNCTION
…
END;//END TYPE BODY
如下所示:
CREATE TYPE HUMAN AS OBJECT(
NAME VARCHAR2(20),
SEX VARCHAR2(1),– F : FEMALE M:MALE
BIRTHDAY DATE,
NOTE VARCHAR2(300),
MEMBER FUNCTION GET_AGE RETURN NUMBER
)
–BODY
CREATE TYPE BODY HUMAN AS
MEMBER FUNCTION GET_AGE RETURN NUMBER AS
V_MONTHS NUMBER;
BEGIN
SELECT FLOOR(MONTHS_BETWEEN(SYSDATE,BIRTHDAY)/12) INTO V_MONTHS FROM DUAL;
RETURN V_MONTHS;
END;
END;
注意:BODY的格式,不是AS OBJECT,也不是用小括号括起来的.MEMBER FUNCTION 后的AS或IS不能省略.
还以STUDENTS表为例(注:如果类型以被某个表使用,是不能修改的,必须把相关的表删除,然后把类型删除,在一个一个新建,这里就省略了,参见前文所述)
SELECT S.STUDENT.GET_AGE() FROM STUDENTS S
在提起注意:表名一定要有别名.GET_AGE()的括号不能省略,否则会提示错误.
下面演示在一个匿名过程中的使用情况:
SET SERVEROUTPUT ON
DECLARE
AA HUMAN;
AGE NUMBER;
BEGIN
AA := HUMAN('xingFairy','M',TO_DATE('19830714','YYYYMMDD'),'过程');
AGE := AA.GET_AGE();
DBMS_OUTPUT.PUT_LINE(AGE);
END;
映射方法:
映射方法是一种不带参数,并返回标准的标量Oracle SQL数据类型的方法,如NUMBER,VARCHAR2,Oracle将间接地使用这些方法执行比较运算.
映射方法最重要的一个特点是:当在WHERE或ORDER BY等比较关系子句中使用对象时,会间接地使用映射方法.
映射方法的声明只过是在普通方法声明的前面加一个 MAP而以,注意:映射方法是一种不带参数的方法.
MAP MEMBER FUNCTION function_name RETURN return_type
修改前文提到的HUMAN类型:
–映射方法 MAP
CREATE TYPE HUMAN AS OBJECT(
NAME VARCHAR2(20),
SEX VARCHAR2(1),– F : FEMALE M:MALE
BIRTHDAY DATE,–注册日期
REGISTERDAY DATE,
NOTE VARCHAR2(300),
MEMBER FUNCTION GET_AGE RETURN NUMBER,
MAP MEMBER FUNCTION GET_GRADE RETURN NUMBER
)
CREATE TYPE BODY HUMAN AS
———————–
MEMBER FUNCTION GET_AGE RETURN NUMBER AS
V_MONTHS NUMBER;
BEGIN
SELECT FLOOR(MONTHS_BETWEEN(SYSDATE,BIRTHDAY)/12) INTO V_MONTHS FROM DUAL;
RETURN V_MONTHS;
END;
————————
MAP MEMBER FUNCTION GET_GRADE RETURN NUMBER AS
BEGIN
RETURN MONTHS_BETWEEN(SYSDATE,BIRTHDAY);
END;
END;
插入数据:
INSERT
INTO STUDENTS (STUDENT) VALUES
(HUMAN('xling','M',TO_DATE('19830714','YYYYMMDD'),TO_DATE('20020915','YYYYMMDD'),'
测试MAP方法'));
INSERT INTO STUDENTS (STUDENT) VALUES
(HUMAN('fairy','M',TO_DATE('19830714','YYYYMMDD'),TO_DATE('20010915','YYYYMMDD'),'
测试MAP方法'));
INSERT INTO STUDENTS (STUDENT) VALUES
(HUMAN('snow','M',TO_DATE('19830714','YYYYMMDD'),TO_DATE('20020915','YYYYMMDD'),'
测试MAP方法'));
在执行上面的操作后,用下面这个SELECT语句可以看出映射方法的效果:
SELECT S.STUDENT.NAME,S.STUDENT.GET_GRADE() FROM STUDENTS S ORDER BY STUDENT
它是按MAP方法GET_GRADE()的值进行排序的.注意是ORDER BY STUDENT,在提起一次需要注意,一定要用表的别名,方法后的括号不能省略,即使没有参数.
如果想以MAP方法的结果为条件,可以如下:
SELECT S.STUDENT.NAME,S.STUDENT.GET_GRADE() FROM STUDENTS S WHERE S.STUDENT.GET_GRADE() > 50
SELECT
S.STUDENT.NAME,S.STUDENT.GET_GRADE() FROM STUDENTS S WHERE STUDENT >
HUMAN(NULL,NULL,NULL,TO_DATE('20020101','YYYYMMDD'),NULL);
排序方法:
先说一下SELF,Oracle里对象的SELF和JAVA里的this是同一个意思.
对象的排序方法具有一个与对象类型相同的参数,暂称为ARG1,用于和SELF对象进行比较.如果调用方法的SELF对象比ARG1小,返回负值,如果相等,返回0,如果SELF大于ARG1,则返回值大于0.
–映射方法 MAP
CREATE TYPE HUMAN AS OBJECT(
NAME VARCHAR2(20),
SEX VARCHAR2(1),– F : FEMALE M:MALE
BIRTHDAY DATE,
REGISTERDAY DATE,
NOTE VARCHAR2(300),
MEMBER FUNCTION GET_AGE RETURN NUMBER,
ORDER MEMBER FUNCTION MATCH(I_STUDENT IN HUMAN) RETURN NUMBER
)
CREATE TYPE BODY HUMAN AS
———————–
MEMBER FUNCTION GET_AGE RETURN NUMBER AS
V_MONTHS NUMBER;
BEGIN
SELECT FLOOR(MONTHS_BETWEEN(SYSDATE,BIRTHDAY)/12) INTO V_MONTHS FROM DUAL;
RETURN V_MONTHS;
END;
————————
ORDER MEMBER FUNCTION MATCH(I_STUDENT IN HUMAN) RETURN NUMBER AS
BEGIN
RETURN REGISTERDAY – I_STUDENT.REGISTERDAY;
END;
END;
注意:在声明的时候,ORDER方法的参数类型要和SELF的类型一致.
SET SERVEROUTPUT ON
DECLARE
S1 HUMAN;
S2 HUMAN;
BEGIN
S1 := HUMAN('xling',NULL,NULL,TO_DATE('20020915','YYYYMMDD'),NULL);
S2 := HUMAN('snow',NULL,NULL,TO_DATE('20010915','YYYYMMDD'),NULL);
IF S1 > S2 THEN
DBMS_OUTPUT.PUT_LINE(S1.NAME);
ELSIF S1 < S2 THEN
DBMS_OUTPUT.PUT_LINE(S2.NAME);
ELSE
DBMS_OUTPUT.PUT_LINE('EQUAL');
END IF;
END;
注意S1 和 S2是怎么比较的.
映射方法具有效率方面的优势,因为它把每个对象与单个标量值联系在一起;排序方法有灵活方面的优势,它可以在两个对象之间进行任意复杂的比较.排序方法比映射方法的速度慢.
转载请注明:学时网 » Oracle对象类型详解