Question

我需要一些帮助来完成这个程序。问题是如何对不同的SELECT进行别名，以便我可以在RIGHT OUTER JOIN上匹配它们。传统的AS不起作用。有什么想法吗？

PROCEDURE "GetProcessedEvents"
(
"In_OccurredFrom" TIMESTAMP,
"In_OccurredTo" TIMESTAMP,
"In_Interval" DECIMAL,
"Out_Cursor" OUT "Ref_Cursor"
)
IS
BEGIN    
OPEN "Out_Cursor" FOR      
  SELECT
    COUNT("Id") AS "ProcessedEvents",
    TO_TIMESTAMP('1900-01-01 00:00:00.000000', 'YYYY-MM-DD HH24:MI:SS.FF') + NUMTODSINTERVAL(FLOOR(DATEDIFF('MI', '1900-01-01', "ProcessedDate")/"In_Interval")*"In_Interval", 'MINUTE') AS "EventDts"
  FROM "Dat_ScheduledEvent"
  RIGHT OUTER JOIN
    (SELECT "Dts" FROM TABLE("F_GetDateIntervalTable"("In_OccurredTo", "In_OccurredFrom", "In_Interval")))
  ON "EventDts" = "Dts"
  WHERE
    "ProcessedDate" BETWEEN "In_OccurredFrom" AND "In_OccurredTo"
  GROUP BY
    TO_TIMESTAMP('1900-01-01 00:00:00.000000', 'YYYY-MM-DD HH24:MI:SS.FF') + NUMTODSINTERVAL(FLOOR(DATEDIFF('MI', '1900-01-01', "ProcessedDate")/"In_Interval")*"In_Interval", 'MINUTE');
END "GetProcessedEvents";

Answer 1

别名列时

AS是可选的，但在别名表（包括子查询）时，Oracle根本不允许这样做。因此，请为子查询指定名称，但不要包含AS关键字：

  FROM "Dat_ScheduledEvent" alias1
  RIGHT OUTER JOIN (
    SELECT "Dts"
    FROM TABLE("F_GetDateIntervalTable"("In_OccurredTo", "In_OccurredFrom",
      "In_Interval"))
  ) alias2

但是你还有其他一些问题。首先，创建具有必须用双引号括起来的标识符的对象是很难的;有混合案例标识符真的值得吗？其次，看起来EventDts只是一个派生字段，并且除了order by子句之外，列别名不能在查询中的任何其他位置使用，因此您可能需要将其作为子查询 - 这意味着您也可以在group by中使用它。第三，DATEDIFF不是内置的Oracle函数 - 除非你自己创建了它，否则你需要使用其他东西。

  SELECT
    COUNT(alias1."Id") AS "ProcessedEvents",
    alias1."EventDts"        
  FROM (
    SELECT "Id",
      TO_TIMESTAMP('1900-01-01 00:00:00.000000', 'YYYY-MM-DD HH24:MI:SS.FF')
        + NUMTODSINTERVAL(FLOOR(DATEDIFF('MI', '1900-01-01', "ProcessedDate")
          /"In_Interval")*"In_Interval", 'MINUTE') AS "EventDts"
    FROM "Dat_ScheduledEvent"
    WHERE "ProcessedDate" BETWEEN "In_OccurredFrom" AND "In_OccurredTo"
  ) alias1
  RIGHT OUTER JOIN (
    SELECT "Dts"
    FROM TABLE("F_GetDateIntervalTable"("In_OccurredTo", "In_OccurredFrom",
      "In_Interval"))
  ) alias2
  ON alias1."EventDts" = alias2."Dts"
  GROUP BY alias2."EventDts";

DATEDIFF仍然需要更换，但我不太清楚你在做什么。如果F_GetDateIntervalTable产生了合适的间隔，我不确定为什么你需要这样做;您是否希望在时间间隔内找到ProcessedDate个值，即>= Dt and < Dt + InInterval？当然，取决于InInterval实际代表的内容。

假设F_GetDateIntervalTable为您提供感兴趣期间内每个区间的开始，您可以这样做：

PROCEDURE "GetProcessedEvents"
(
"In_OccurredFrom" TIMESTAMP,
"In_OccurredTo" TIMESTAMP,
"In_Interval" DECIMAL,
"Out_Cursor" OUT "Ref_Cursor"
)
IS
BEGIN    
OPEN "Out_Cursor" FOR
  SELECT
    COUNT(Event."Id") AS "ProcessedEvents",
    DateInt."Dts"
  FROM (
    SELECT "Dts", COALESCE(LEAD("Dts")
      OVER (ORDER BY "Dts"), "In_OccurredTo") as "NextDts"
    FROM TABLE("F_GetDateIntervalTable"("In_OccurredTo", "In_OccurredFrom",
      "In_Interval"))
  ) DateInt
  LEFT JOIN "Dat_ScheduledEvent" Event
  ON Event."ProcessedDate" >= DateInt."Dts"
  AND Event."ProcessedDate" < DateInt."NextDts"
  GROUP BY DateInt."Dts"
  ORDER BY DateInt."Dts";
END "GetProcessedEvents";
/

使用LEAD()可以查看下一个结果，以便子查询将时间戳列表转换为from-和to-timestamps列表（最后一个结束于In_OccurredTo），这允许您在数据表中查找属于这些范围的匹配记录 - 不需要操纵ProcessedDate。

有一些假数据：

create table "Dat_ScheduledEvent" ("Id" number, "ProcessedDate" timestamp);

insert into "Dat_ScheduledEvent" ("Id", "ProcessedDate")
  values (1, to_timestamp('2013-02-06 09:45', 'YYYY-MM-DD HH24:MI'));
insert into "Dat_ScheduledEvent" ("Id", "ProcessedDate")
  values (1, to_timestamp('2013-02-06 09:50', 'YYYY-MM-DD HH24:MI'));
insert into "Dat_ScheduledEvent" ("Id", "ProcessedDate")
  values (1, to_timestamp('2013-02-06 10:15', 'YYYY-MM-DD HH24:MI'));

...并从SQL * Plus调用该过程：

var r refcursor;
alter session set nls_date_format = 'YYYY-MM-DD HH24:MI';
exec "GetProcessedEvents"(to_date('2013-02-06 08:00'), to_date('2013-02-06 12:00'), 1/24, :r);
print :r

......我明白了：

ProcessedEvents        Dts                       
---------------------- ------------------------- 
0                      06-FEB-13 08.00.00.000000000 
2                      06-FEB-13 09.00.00.000000000 
1                      06-FEB-13 10.00.00.000000000 
0                      06-FEB-13 11.00.00.000000000

Answer 2

是的，我有一些想法。使用 WITH子句可能很有用，它允许您将复杂的SELECT语句分解为子查询，从而有效地改进代码结构，当然，您可以通过别名引用它们（子查询）。

以下是您使用的一个小例子：

WITH subquery AS (SELECT sysdate FROM dual)
-- Here goes main query where subquery must be used
SELECT *
FROM subquery


OPEN "Out_Cursor" FOR     
WITH Q AS (SELECT "Dts" FROM TABLE("F_GetDateIntervalTable"("In_OccurredTo", "In_OccurredFrom", "In_Interval")))

SELECT
    COUNT("Id") AS "ProcessedEvents",
    TO_TIMESTAMP('1900-01-01 00:00:00.000000', 'YYYY-MM-DD HH24:MI:SS.FF') + NUMTODSINTERVAL(FLOOR(DATEDIFF('MI', '1900-01-01', "ProcessedDate")/"In_Interval")*"In_Interval", 'MINUTE') AS "EventDts"
  FROM "Dat_ScheduledEvent"
  RIGHT OUTER JOIN
    Q
  ON "EventDts" = "Dts"
  WHERE
    "ProcessedDate" BETWEEN "In_OccurredFrom" AND "In_OccurredTo"
  GROUP BY
    TO_TIMESTAMP('1900-01-01 00:00:00.000000', 'YYYY-MM-DD HH24:MI:SS.FF') + NUMTODSINTERVAL(FLOOR(DATEDIFF('MI', '1900-01-01', "ProcessedDate")/"In_Interval")*"In_Interval", 'MINUTE');
END "GetProcessedEvents";

Answer 3

尝试删除变量名称旁边的双引号，除非语法要求它们。这可能会让解析器感到困惑。

Oracle过程 - 使用子查询加入

3 个答案: