Question

declare
  n_docid number;
  n_orgid number;
  n_docstateid number;
  n_docstateversion number;
  n_doctypeid number;
  n_count number;
begin
  for I in (select DOCID from DC_EXP_BO where INF_DATEDOCUMENT > '01.01.17' and 
INF_DATEDOCUMENT < '31.12.17')
loop
begin
  select DOCID, DOCSTATEID, DOCTYPEID, DOCSTATE_VERSION into 
  n_docid, n_docstateid, n_doctypeid, n_docstateversion from DOC 
  where DOCID = I.DOCID;

  select ORGID into n_orgid from ORG
  where systemname = (select distinct RB_CODEGRBS from DC_EXP_BO where DOCID = n_docid);


  select count(*) into n_count from ROUTECONTEXT
  where DOCID = n_docid and ORGID = n_orgid;

  if (n_count = 0) then
  insert into ROUTECONTEXT 
    (ROUTECONTEXTID, VERSION, DOCID, LOCALDOCSTATEID, OWNERID, LASTPRINTDATE, PRINTED, RECEIVED, ORGID, ARCHIVE, EXPORTSTATUS, DOCTYPEID, DOCSTATEID, DOCSTATE_VERSION, DELETED)
  values 
    (sq_routeContext.NEXTVAL, 0, n_docid, n_docstateid, null, null, 0, 0, n_orgid, 0, 'NOT_EXPORTED', n_doctypeid, n_docstateid, n_docstateversion, 0);
  end if;
exception
  when no_data_found then
continue;
end;
end loop;
end;
/

我们为生产编写了datafix。而且有一个问题。对于语句select，可以返回大约100万个ID。是否有可能优化此查询？

Answer 1

您可以进行的最有效的优化是使用单个基于集合的操作替换循环中的逐行插入和查找：

insert into ROUTECONTEXT 
    (ROUTECONTEXTID, VERSION, DOCID, LOCALDOCSTATEID, OWNERID, LASTPRINTDATE, PRINTED, RECEIVED, ORGID, ARCHIVE, EXPORTSTATUS, DOCTYPEID, DOCSTATEID, DOCSTATE_VERSION, DELETED)
select sq_routeContext.NEXTVAL, 
          0, 
          DOC.DOCID, 
          DOC.DOCSTATEID, 
          null,
          null,
          0,
          0,
          org.ORGID,
          0,
          'NOT_EXPORTED',
          DOC.DOCSTATEID, 
          DOC.DOCTYPEID, 
          DOC.DOCSTATE_VERSION ,
          0
from DC_EXP_BO
    join doc
        on DC_EXP_BO.DOCID = DOC.DOCID 
    join org 
        on org.systemname =  DC_EXP_BO.RB_CODEGRBS 
where DC_EXP_BO.INF_DATEDOCUMENT > date '2017.01.01'
and DC_EXP_BO.INF_DATEDOCUMENT < date '2017.12.31' 
and not exists ( select null 
                 from ROUTECONTEXT
                 where ROUTECONTEXT.DOCID = doc.docid 
                 and ROUTECONTEXT.ORGID = org.orgid 
               )
/

假设DC_EXP_BO.INF_DATEDOCUMENT是DATE数据类型，使用正确的日期语义可能会更具性能，并且肯定会更安全。但它是一个字符串，用您在问题中发布的内容替换WHERE子句中的那一部分。

有没有办法优化此查询？

1 个答案: