这是我们购买的产品的情景。该产品允许我们创建自定义字段，但这些字段存储为CUSTOM表中的ROWS。
我想编写一个查询，该查询将连接到多个自定义字段并获取单行。
让我举个例子。
1) PERSON TABLE (ID int, NAME varchar2(30));
2) CUSTOMFIELDS TABLE(CFID int, CFTable varchar2(30), CFFieldName varchar2(30), CFFieldType varchar2(30));
3) CUSTOMFIELDVALUE TABLE(CFVID int, CFID int, CFFieldName varchar2(100), CFFieldValue varchar2(100));
我的人员表有一条记录
1) 1001 - Clark Kent
假设我为名为AGE和WEIGHT的人创建了两个自定义字段。 在这种情况下，将在CUSTOMFIELDS表中创建两条记录。
1) 100 - PERSON - AGE - INTEGER
2) 200 - PERSON - WEIGHT - INTEGER
现在这些自定义字段的值将像这样存储在CUSTOMFIELDVALUE表中。
1) 100 - 100 - 1001 - 44
2) 101 - 200 - 1001 - 200 lbs
我想写一个选择查询来获取像这样的记录
PERSON, AGE , WEIGHT
Clark Kent, 44, 200 lbs
我在想如何通过纯SQL实现这一目标。自定义字段的数量可以增加或减少，具体取决于产品的配置。
答案 0 :(得分：0)
或许这样的事情：
select p.name, age.CFFieldValue, weight.CFFieldValue
from person p
inner join CUSTOMFIELDVALUE age
on p.id = age.CFFieldName
inner join CUSTOMFIELDS age_f
on age_f.cfid = age.cfid and age_f.CFFieldName = 'AGE' and age_f.CFTable = 'PERSON'
inner join CUSTOMFIELDVALUE weight
on p.id = weight.CFFieldName and weight_f.CFFieldName = 'AGE' and weight_f.CFTable = 'PERSON'
inner join CUSTOMFIELDS weight_f
on weight_f.cfid = weight.cfid;
我不完全清楚person.id是否加入customfieldvalue.cfieldname，但你确实说100 - 100 - 1001 - 44所以我猜你试图说1001是第三列吗？本质上，查询过滤customfieldvalues表并将其用作两个单独的表。我认为这就是你想要的 - 你的问题并不是很清楚。
答案 1 :(得分：0)
这是一个有趣的问题。您想要动态更改列的数量和名称。使用“普通”SQL无法创建。我尝试使用PIPELINED FUNCTION创建样本。
我首先创建了表：
CREATE TABLE PERSON (ID int, NAME varchar2(30));
CREATE TABLE CUSTOMFIELDS (CFID int, CFTable varchar2(30), CFFieldName varchar2(30), CFFieldType varchar2(30));
CREATE TABLE CUSTOMFIELDVALUE (CFVID int, CFID int, CFFieldName varchar2(100), CFFieldValue varchar2(100));
INSERT INTO PERSON(id, name) values(1001, 'Clark Kent');
INSERT INTO CUSTOMFIELDS(CFID, CFTable, CFFieldName, CFFieldType) values(100, 'PERSON', 'AGE', 'INTEGER');
INSERT INTO CUSTOMFIELDS(CFID, CFTable, CFFieldName, CFFieldType) values(200, 'PERSON', 'WEIGHT', 'INTEGER');
......我放了一些数据：
INSERT INTO CUSTOMFIELDVALUE (CFVID, CFID, CFFieldName, CFFieldValue) values(100, 100, 1001, 44);
INSERT INTO CUSTOMFIELDVALUE (CFVID, CFID, CFFieldName, CFFieldValue) values(101, 200, 1001, 200);
然后我创建了一个对象类型：
CREATE TYPE CustomFieldType AS OBJECT
(
row_id number,
fieldType varchar2(200),
person_id number,
fieldValue1 varchar2(2000),
fieldValue2 varchar2(2000),
fieldValue3 varchar2(2000),
fieldValue4 varchar2(2000),
fieldValue5 varchar2(2000)
)
/
CREATE TYPE CustomFieldTypeSet AS TABLE OF CustomFieldType
/
还创建了流水线功能：
CREATE OR REPLACE
FUNCTION GET_PERSON_FIELDS(person_id_in IN NUMBER
,field_names_in IN VARCHAR2) RETURN CustomFieldTypeSet
PIPELINED
IS
-- constructor CustomFieldType()
l_header_row CustomFieldType := CustomFieldType(NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL);
l_data_row CustomFieldType := CustomFieldType(NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL);
l_tablen BINARY_INTEGER;
l_tab DBMS_UTILITY.uncl_array;
l_num_of_field_values PLS_INTEGER := 5;
l_counter PLS_INTEGER := 1;
l_position PLS_INTEGER;
l_field_names_in VARCHAR2(2000) := field_names_in;
TYPE type_header_hash IS TABLE OF PLS_INTEGER INDEX BY VARCHAR2(200);
l_header_hash type_header_hash;
BEGIN
-- 1) check, what fields you can display
IF (l_field_names_in IS NULL) THEN
<<get_all_fields>>
FOR cur_all_fields IN (SELECT DISTINCT flds.CFFIELDNAME
FROM CUSTOMFIELDS flds
,CUSTOMFIELDVALUE cfv
WHERE cfv.CFID = flds.CFID
AND flds.CFTable = 'PERSON' ) LOOP
l_field_names_in := l_field_names_in ||
cur_all_fields.CFFIELDNAME ||
',';
END LOOP get_all_fields;
END IF;
-- 2) generate header (function RTRIM prevent ORA-00931 exception!)
DBMS_UTILITY.comma_to_table(list => RTRIM(l_field_names_in, ','), tablen => l_tablen, tab => l_tab);
l_header_row.row_id := 1;
l_header_row.fieldType := 'HEADER';
<<header_cursor>>
FOR i IN 1..l_tablen LOOP
IF (i = 1) THEN
l_header_row.fieldValue1 := l_tab(i);
l_header_hash(l_tab(i)) := i;
ELSIF (i = 2) THEN
l_header_row.fieldValue2 := l_tab(i);
l_header_hash(l_tab(i)) := i;
ELSIF (i = 3) THEN
l_header_row.fieldValue3 := l_tab(i);
l_header_hash(l_tab(i)) := i;
ELSIF (i = 4) THEN
l_header_row.fieldValue4 := l_tab(i);
l_header_hash(l_tab(i)) := i;
ELSIF (i = 5) THEN
l_header_row.fieldValue5 := l_tab(i);
l_header_hash(l_tab(i)) := i;
END IF;
END LOOP header_cursor;
-- 3) print data to SQL (over pipe)...
PIPE ROW(l_header_row);
FOR cur_persons IN (SELECT ID
FROM PERSON
WHERE ID = COALESCE(person_id_in, ID) ) LOOP
l_data_row.row_id := NULL;
l_data_row.person_id := NULL;
l_data_row.fieldType := NULL;
l_data_row.fieldValue1 := NULL;
l_data_row.fieldValue2 := NULL;
l_data_row.fieldValue3 := NULL;
l_data_row.fieldValue4 := NULL;
l_data_row.fieldValue5 := NULL;
l_data_row.fieldType := 'DATA';
FOR cur_data IN (SELECT p.ID AS person_id
,cfv.CFID
,flds.CFTABLE
,flds.CFFIELDNAME
,cfv.CFFIELDVALUE
FROM PERSON p
,CUSTOMFIELDS flds
,CUSTOMFIELDVALUE cfv
WHERE p.ID = cur_persons.ID
AND p.ID = cfv.CFFIELDNAME
AND cfv.CFID = flds.CFID ) LOOP
l_data_row.person_id := cur_persons.ID;
l_position := NULL;
IF (l_header_hash.EXISTS(cur_data.CFFIELDNAME)) THEN
l_position := l_header_hash(cur_data.CFFIELDNAME);
END IF;
IF (l_position = 1) THEN
l_data_row.fieldValue1 := cur_data.CFFIELDVALUE;
ELSIF (l_position = 2) THEN
l_data_row.fieldValue2 := cur_data.CFFIELDVALUE;
ELSIF (l_position = 3) THEN
l_data_row.fieldValue3 := cur_data.CFFIELDVALUE;
ELSIF (l_position = 4) THEN
l_data_row.fieldValue4 := cur_data.CFFIELDVALUE;
ELSIF (l_position = 5) THEN
l_data_row.fieldValue5 := cur_data.CFFIELDVALUE;
END IF;
END LOOP;
l_counter := l_counter + 1;
l_data_row.row_id := l_counter;
PIPE ROW(l_data_row);
END LOOP;
RETURN;
END GET_PERSON_FIELDS;
您可以使用SQL获取示例数据（注意：防止异常ORA-22905，您必须设置会话变量“ALTER SESSION SET CURSOR_SHARING = EXACT;”）：
SELECT * FROM TABLE(GET_PERSON_FIELDS(1001,'AGE,WEIGHT'));
这是输出：
ROW_ID FIELDTYPE PERSON_ID FIELDVALUE FIELDVALUE FIELDVALUE
------ ---------- --------- ---------- ---------- ----------
1 HEADER AGE
2 DATA 1001 44
在第一列是标题，其中存储有关字段名称的信息，并且在标题之后存储数据。您可以使用这些SQL的组合：
SELECT * FROM TABLE(GET_PERSON_FIELDS(1001,'AGE,WEIGHT'));
SELECT * FROM TABLE(GET_PERSON_FIELDS(1002,'AGE,GENDER'));
SELECT * FROM TABLE(GET_PERSON_FIELDS(1001,NULL));
SELECT * FROM TABLE(GET_PERSON_FIELDS(NULL,NULL));
最后，当我输入一些样本数据时，如下：
INSERT INTO PERSON(id, name) values(1002, 'Lois Lane');
INSERT INTO CUSTOMFIELDS(CFID, CFTable, CFFieldName, CFFieldType) values(300, 'PERSON', 'GENDER', 'VARCHAR');
INSERT INTO CUSTOMFIELDS(CFID, CFTable, CFFieldName, CFFieldType) values(400, 'PERSON', 'SINGLE', 'VARCHAR');
INSERT INTO CUSTOMFIELDVALUE (CFVID, CFID, CFFieldName, CFFieldValue) values(102, 100, 1002, 45);
INSERT INTO CUSTOMFIELDVALUE (CFVID, CFID, CFFieldName, CFFieldValue) values(103, 300, 1002, 'FEMALE');
INSERT INTO CUSTOMFIELDVALUE (CFVID, CFID, CFFieldName, CFFieldValue) values(104, 400, 1002, 'YES');
...并运行此SQL命令：
SELECT * FROM TABLE(GET_PERSON_FIELDS(NULL,NULL));
...输出看起来像这样：
ROW_ID FIELDTYPE PERSON_ID FIELDVALUE FIELDVALUE FIELDVALUE FIELDVALUE
------ ---------- --------- ---------- ---------- ---------- ----------
1 HEADER AGE GENDER SINGLE WEIGHT
2 DATA 1001 44 200
3 DATA 1002 45 FEMALE YES