我有一个表,它有选择查询和插入存储为varchars的查询。我必须执行select查询并使用过程使用insert查询插入select查询的结果。现在我立即执行并将select查询批量收集到varchars表中。
在此之后,我转到vartable的每一行并获取值并将其放入insert查询中。我有许多对多个表的select和insert查询对。所以这必须动态完成。 我的问题是,有没有更好的方法来存储选择查询的结果?除了使用varchar表?因为select查询的结果集可能包含数百万条记录,这可能会导致问题。 使用嵌套表格类型并使用 EXTEND 解决问题吗?
PROCEDURE SEL_INS
AS
CURSOR C
IS
SELECT
SELEQRY SELQRY,
INSQUERY INSERTQRY,
cols COLS
FROM TAB1;
selqry VARCHAR2(1000);
insqry VARCHAR2(1000);
tab1 vartable:=vartable();
cols NUMBER;
BEGIN
tab1:=vartable(NULL);
FOR X IN C
LOOP
selqry:= X.SELQRY;
cols:=X.COLS;
EXECUTE immediate selqry bulk collect INTO tab1;
-- select statement is concatenated before executing. so that each index has one record
--with values separated by commas
--- a sample column in tab1 will have values like (abc,abc1,abc2)
FOR i IN 1..tab1.count
LOOP
insqry :=X.INSERTQRY;
--- insert query will have values like insert into tab2 values('abc,'abc1','abc2')
EXECUTE immediate insqry;
END LOOP;
END LOOP;
END SEL_INS;
vartable是varchars2(4000)
类型的表答案 0 :(得分:3)
如评论中所述,您应尝试将您的陈述重写为INSERT INTO ... SELECT ...
。让我们假设无论出于何种原因这是不可能的。在这种情况下,您可以使用以下程序:
PROCEDURE SEL_INS AS
CURSOR C IS
SELECT SELEQRY, INSQUERY, COLS
FROM TAB1;
selqry VARCHAR2(1000);
insqry VARCHAR2(1000);
tab1 vartable;
cols NUMBER;
cur INTEGER;
res INTEGER;
col_cnt INTEGER;
desctab DBMS_SQL.DESC_TAB;
i INTEGER;
BEGIN
FOR aQuery IN C LOOP
tab1 := vartable();
cur := DBMS_SQL.OPEN_CURSOR;
DBMS_SQL.PARSE(cur, aQuery.SELEQRY, DBMS_SQL.NATIVE);
DBMS_SQL.DESCRIBE_COLUMNS(cur, col_cnt, desctab)
FOR i IN 1..col_cnt LOOP
DBMS_SQL.DEFINE_COLUMN(cur, i, desctab(i).COL_NAME, 2000);
END LOOP;
res := DBMS_SQL.EXECUTE_AND_FETCH(cur, TRUE);
FOR i IN 1..col_cnt LOOP
tab1.EXTEND;
DBMS_SQL.COLUMN_VALUE(cur, i, tab1(tab1.LAST));
END LOOP;
DBMS_SQL.CLOSE_CURSOR(cur);
-- ... do whatever with tab1(xyz) -> otherwise this procedure would be an overkill
cur := DBMS_SQL.OPEN_CURSOR;
DBMS_SQL.PARSE(cur, aQuery.INSQUERY, DBMS_SQL.NATIVE);
i := tab1.FIRST;
WHILE i IS NOT NULL LOOP
DBMS_SQL.BIND_VARIABLE(cur, ':b'||i, tab1(i));
i := tab1.NEXT(i);
END LOOP;
res := DBMS_SQL.EXECUTE(cur);
DBMS_SQL.CLOSE_CURSOR(cur);
END LOOP;
END;
注意,此过程假设所有列都是VARCHAR2
数据类型(最大长度为2000个字符)。如果您有其他数据类型,则必须将行DBMS_SQL.DEFINE_COLUMN(cur, i, desctab(i).COL_NAME, 2000);
扩展为IF desctab(c).col_type = 1 THEN ...
另外,注意DBMS_SQL.EXECUTE_AND_FETCH
将失败,除非您的select只返回一行。如果您的查询可能只返回一行,则必须使用
DBMS_SQL.EXECUTE(cur);
WHILE (DBMS_SQL.FETCH_ROWS(cur) > 0) LOOP
...
END LOOP;
请参阅Oracle Built-in Data Types以获取每种数据类型的代码编号。
答案 1 :(得分:1)
假设您能够更改存储在要运行的sql语句的表中的数据,这是我获取存储数据的示例:
INSERT INTO sel_ins (selqry, insquery)
SELECT 'select col1, col2, col3 from table1' selqry, 'insert into other_table1 (col1, col2, col3)') insqry FROM dual
UNION ALL
SELECT 'select col1, col2 from table2' selqry, 'insert into other_table2 (col1, col2)') insqry FROM dual
UNION ALL
SELECT 'select col1, col2, col3 from table3 where col4 = ''fred''' selqry, 'insert into other_table3 (col1, col2, col3)') insqry FROM dual;
通过这样做,您执行插入的过程现在变得更加简单:
PROCEDURE sel_ins IS
CURSOR ins_sel_cur IS
SELECT seleqry selqry, insquery insqry, cols cols
FROM tab1;
v_selqry tab1.seleqry%TYPE;
v_insqry tab1.insquery%TYPE;
BEGIN
FOR ins_sel_rec IN ins_sel_cur
LOOP
EXECUTE IMMEDIATE ins_sel_rec.v_insqry || CHR(10) || ins_sel_rec.v_selqry;
END LOOP;
END;
/
这样,您就不会从表中检索大量数据并将其存储在内存中,只是为了获取该数据并将其逐行添加回另一个表中 - 您正在执行此操作所有这些都在一个DML声明中工作(相当于一个建造者让卡车包含她需要提供的所有砖块,直到她建造墙壁,而不是在驱动器的底部,然后去并逐一取出每块砖。这应该会使事情变得更快,更不用说更容易阅读,维护等。
如果您希望单独存储列,例如:
INSERT INTO sel_ins (selqry, insquery, cols)
SELECT 'select <COLS> from table1' selqry, 'insert into other_table1 (<COLS>)') insqry, 'col1, col2, col3' cols FROM dual
UNION ALL
SELECT 'select <COLS> from table2' selqry, 'insert into other_table2 (<COLS>)') insqry, 'col1, col2' cols FROM dual
UNION ALL
SELECT 'select <COLS> from table3 where col4 = ''fred''' selqry, 'insert into other_table3 (<COLS>)') insqry, 'col1, col2, col3' cols FROM dual;
然后你的执行立即成为:
execute immediate replace(ins_sel_rec.v_insqry, '<COLS>', ins_sel_rec.cols) ||
chr(10) ||
replace(ins_sel_rec.v_selqry, '<COLS>', ins_sel_rec.cols);