oracle plsql对行进行分块

Question

I have a select statement which returns 0 or more rows.

I'm trying to come up with a plsql proc with a cursor to produce xml output fro, all rows returned into 100 rows at a time. i'm doing this to chunk loo rows at a time based on requirement.

So basically my program should follow below logic

cursor c1 is select id,code_id,code_desc from table order by id; --returns some rows  

if c1%notfound
  then return;` -- exit from procedure 
else
loop
  grab first 100 rows from select and append to a variable
  and assign it to a variable; 
  update this variable into a clob field in a table. 
  grab next 100 rows and append into a variable
  update this variable into a clob field in a table in another row;see below 
  table data 

   and so on
   and grab remaining rows and append into a variable 
   print the variable;
until no data found;
exit

I'm trying to do convert the output from select statement into xml text.

The output should look something like below:

TABLE: STG_XML_DATA

LOOP_NO(NUMBER), XML_TEXT(CLOB), ROWS_PROCESSED
1                <XML><id>1</ID><id>2</ID>..<ID>100</ID></XML>     100
2                <XML><id>101</ID><id>102</ID>..<ID>200</ID></XML> 200 
3                <XML><id>301</ID><id>102</ID>..<ID>320</ID></XML> 20

Can someone please help

Answer 1

First of all, can you do this with a single INSERT ... SELECT statement that does what you want with reasonable performance? If you're doing a million rows, yes, breaking them up into chunks may be a good idea. But if it's 100, that might be your best bet.

For your actual question, you want to use BULK COLLECT into a collection variable and possibly FORALL . So your function is going to look something like this:

DECLARE
    TYPE id_tt IS TABLE OF NUMBER;
    TYPE desc_tt IS TABLE OF VARCHAR2(100);
    l_ids         id_tt;
    l_code_ids    id_tt;
    l_code_descs  desc_tt;
    cursor c1 is select id,code_id,code_desc from table 
BEGIN
    OPEN c1;
    LOOP
        FETCH c1 BULK COLLECT INTO l_ids, l_code_ids, l_code_descs
        LIMIT 100;

        EXIT WHEN l_ids.COUNT = 0;

        FORALL idx IN 1..l_ids.COUNT
        INSERT [... some insert statement here ...]

        [... maybe some other processing here...]
    END LOOP;
    CLOSE c1;
END;

What you absolutely do not want to do is fetch a row, process it, fetch another row, etc. SQL is a set-oriented language, so try to operate on sets. Every time you switch context from SQL to PL/SQL there is a cost and it can kill your performance.

See: http://www.oracle.com/technetwork/issue-archive/2012/12-sep/o52plsql-1709862.html