如何解析 varchar2 并将其分成多行？

Question

这是这个问题的后续问题：How do I need to change my sql to get what I want in this case?

我的两张表如下：

      Table 1
id  id2    date
1    2   2015-01-10
2    5   2015-06-13
3    9   2015-09-05
4    10  2015-02-11
5    26  2015-01-10
6    65  2015-01-25

      Table 2
id  id2    data(varchar2)
1    2       A
2    5       A
3    9       A
4    10      B
5    26      B
6    65      B

Table 2 中的数据是 varchar2 类型，其中包含N 数字，由, 分隔，其中N 可能依赖于id2。例如，A 可能是这样的：

1.0,1.1,1.2,1.3,1.4,1.5,2.6,2.7,2.8,2.9, ...(ommitted)..., 9.5,9.9

我想编写一个查询，返回 Table 2 中唯一的 data，它在 Table 1 中具有最大日期，对于上表来说如下：

id2    date         number
2   2015-01-10        1.0
2   2015-01-10        1.1
2   2015-01-10        1.2
2   2015-01-10        1.3
      ...
2   2015-01-10        9.5
2   2015-01-10        9.9
10  2015-02-11        ***
10  2015-02-11        ***
      ...
10  2015-02-11        ***

每个唯一的id2 在查询输出中出现N 次。

根据 Fuzzy 的回答，我可以得到唯一的 id2s，如下所示：

select * from (
    select 
        t2.id2, t1.date, t2.data, 
        row_number() over (partition by t2.data order by t1.date desc) rn
    from table1 t1
    join table2 t2 on t1.id2 = t2.id2
) t where rn = 1;

但我不知道如何从那里继续。非常感谢。

Answer 1

您可以使用管道功能：

CREATE or replace TYPE test_type AS TABLE OF varchar2(40)

CREATE or replace FUNCTION test_func (d VARCHAR2)
   RETURN test_type
   PIPELINED
IS
BEGIN
   FOR C1 IN (    SELECT REGEXP_SUBSTR (d, '[^,]+', 1, LEVEL) x
                    FROM DUAL
              CONNECT BY REGEXP_SUBSTR (d, '[^,]+', 1, LEVEL) IS NOT NULL)
   LOOP
      PIPE ROW (c1.x);
   END LOOP;
END;
/

WITH test
     AS ( your_query_here )
SELECT x.id2, x.mydate, y.*
  FROM test x, TABLE (test_func (x.d)) y

我在这里使用 mydate 代替 date（保留字）和 d 代替数据。示例：

WITH test
     AS (SELECT 2 AS id2, SYSDATE AS mydate, '1.0,1.1,1.2,11,1.4,1.5,2.6,2.7,2.8,2.9,44,55' AS d FROM DUAL
         UNION ALL
         SELECT 3 AS id2, SYSDATE + 1 AS mydate, '19.5,19.9,11.5,11.1,21.2,33,1.4,1.5,2.6,2.7,2.8,2.9' AS d
           FROM DUAL
         UNION ALL
         SELECT 4 AS id2, SYSDATE + 1 AS mydate, '9.5,9.9,1.5,1.1,1.2,66,1.4,1.5,2.6,2.7,2.8,2.9' AS d
           FROM DUAL)
SELECT x.id2, x.mydate, y.*
  FROM test x, TABLE (test_func (x.d)) y

部分链接：

How to split comma separated string and pass to IN clause of select statement

Using Pipelined and Parallel Table Functions

Answer 2

正如其他人所说，如果桌子设计得当，您就不必这样做。如果这是一个经常使用的对象，最好解决真正的问题。但是，如果这只是一次性的事情，您可以使用递归查询。这在 Teradata 中有效，Oracle 可能有类似的类似物。

WITH RECURSIVE processDelimited(ID2,dataVarchar2,parsedElement,indexElement) AS (
    SELECT 
     ID2
    ,dataVarchar2
    ,strtok(dataVarchar2,',',1)
    ,0
    FROM tableWithDelimitedColumns
    UNION ALL
    SELECT 
     ID2
    ,dataVarchar2
    ,strtok(dataVarchar2,',',indexElement + 1)    
    ,indexElement + 1
    FROM processDelimited
    WHERE indexElement < 120
    )

SELECT 
 processDelimited.ID2
,parsedElement
,indexElement
FROM processDelimited
INNER JOIN (
    SELECT 
     ID2
    ,DATE_COL
    FROM table1
    QUALIFY row_number() OVER(PARTITION BY ID2 ORDER BY DATE_COL DESC) = 1) AS id2MaxDate
    ON processDelimited.ID2 = id2MaxDate.ID2
WHERE parsedElement IS NOT NULL
ORDER BY processDelimited.ID2

基本上，您只需从空字符串开始，然后循环添加行。

使用的 TD 函数只是从分隔字段中提取索引。例如。

strtok('1,2,3,4,5',',',2) = 2

当然，这在计算上非常昂贵，因此对于大表来说并不是很好。

Answer 3

Oracle 设置：

CREATE TYPE VARCHAR2_TABLE AS TABLE OF VARCHAR2(4000);
/

CREATE OR REPLACE FUNCTION split_String(
  i_str    IN  VARCHAR2,
  i_delim  IN  VARCHAR2 DEFAULT ','
) RETURN VARCHAR2_TABLE DETERMINISTIC
AS
  p_result       VARCHAR2_TABLE := VARCHAR2_TABLE();
  p_start        NUMBER(5) := 1;
  p_end          NUMBER(5);
  c_len CONSTANT NUMBER(5) := LENGTH( i_str );
  c_ld  CONSTANT NUMBER(5) := LENGTH( i_delim );
BEGIN
  IF c_len > 0 THEN
    p_end := INSTR( i_str, i_delim, p_start );
    WHILE p_end > 0 LOOP
      p_result.EXTEND;
      p_result( p_result.COUNT ) := SUBSTR( i_str, p_start, p_end - p_start );
      p_start := p_end + c_ld;
      p_end := INSTR( i_str, i_delim, p_start );
    END LOOP;
    IF p_start <= c_len + 1 THEN
      p_result.EXTEND;
      p_result( p_result.COUNT ) := SUBSTR( i_str, p_start, c_len - p_start + 1 );
    END IF;
  END IF;
  RETURN p_result;
END;
/

CREATE TABLE Table1 ( id, id2, "date" ) AS
SELECT 1,    2,   DATE '2015-01-10' FROM DUAL UNION ALL
SELECT 2,    5,   DATE '2015-06-13' FROM DUAL UNION ALL
SELECT 3,    9,   DATE '2015-09-05' FROM DUAL UNION ALL
SELECT 4,    10,  DATE '2015-02-11' FROM DUAL UNION ALL
SELECT 5,    26,  DATE '2015-01-10' FROM DUAL UNION ALL
SELECT 6,    65,  DATE '2015-01-25' FROM DUAL;

CREATE TABLE Table2 ( id, id2, data ) AS
SELECT 1,    2,       '1.0,1.1,1.2' FROM DUAL UNION ALL
SELECT 2,    5,       '1.0,1.1,1.2' FROM DUAL UNION ALL
SELECT 3,    9,       '1.0,1.1,1.2' FROM DUAL UNION ALL
SELECT 4,    10,      '3.2,3.3,4.5,6.7' FROM DUAL UNION ALL
SELECT 5,    26,      '3.2,3.3,4.5,6.7' FROM DUAL UNION ALL
SELECT 6,    65,      '3.2,3.3,4.5,6.7' FROM DUAL;

查询：

SELECT t.id,
       t.id2,
       t."date",
       d.column_value AS data
FROM   (
  SELECT MAX( t1.id  ) KEEP ( DENSE_RANK LAST ORDER BY t1."date" ) AS id,
         MAX( t1.id2 ) KEEP ( DENSE_RANK LAST ORDER BY t1."date" ) AS id2,
         MAX( t1."date" ) AS "date",
         t2.data
  FROM   Table1 t1
         INNER JOIN
         Table2 t2
         ON ( t1.id = t2.id AND t1.id2 = t2.id2 )
  GROUP BY t2.data
) t,
TABLE( SPLIT_STRING( t.data, ',' ) ) d;

输出：

        ID        ID2 date                DATA
---------- ---------- ------------------- ----
         3          9 2015-09-05 00:00:00 1.0  
         3          9 2015-09-05 00:00:00 1.1  
         3          9 2015-09-05 00:00:00 1.2  
         4         10 2015-02-11 00:00:00 3.2  
         4         10 2015-02-11 00:00:00 3.3  
         4         10 2015-02-11 00:00:00 4.5  
         4         10 2015-02-11 00:00:00 6.7