Oracle 中的子字符串查询 SQL

Question

我有一个 table，其中有 2 列，格式如下。我想从 Col1 中获取所有字符串，这些字符串从 col2 的单词末尾开始作为单独的列。

例如：如果 Col1 有 ABC:EFG:MNO:XYZ 并且 Col2 有 MNO，那么输出将有 L1 作为 XYZ（数据在 Col2 的字符串和 Col1)

的结尾之间

COL1	COL2
ABC:EFG:MNO:XYZ	MNO
PQR:NOM:XYN:SDF:RST:EDF	NOM

我想要低于输出。

L1	L2	L3	L4	L5
XYZ
XYN	SDF	RST	EDF

我尝试使用子字符串，但它对我不起作用。

此致

Answer 1

按照您的说法，这样的查询就可以完成工作。阅读代码中的注释。

SQL> with
  2  test (col1, col2) as
  3    -- sample data; you already have that in your table; don't type it
  4    (select 'ABC:EFG:MNO:XYZ'        , 'MNO' from dual union all
  5     select 'PQR:NOM:XYN:SDF:RST:EDF', 'NOM' from dual
  6    ),
  7  remainder as
  8    -- fetch part of COL1 that follows the COL2 value
  9    (select col1, col2,
 10       substr(col1, instr(col1, col2) + length(col2) + 1) col
 11     from test
 12    )
 13  -- finally, extract up to 5 "words" (as your example suggests) from COL
 14  select regexp_substr(col, '\w+', 1, 1) l1,
 15         regexp_substr(col, '\w+', 1, 2) l2,
 16         regexp_substr(col, '\w+', 1, 3) l3,
 17         regexp_substr(col, '\w+', 1, 4) l4,
 18         regexp_substr(col, '\w+', 1, 5) l5
 19  from remainder;

L1    L2    L3    L4    L5
----- ----- ----- ----- -----
XYZ
XYN   SDF   RST   EDF

SQL>

Answer 2

您可以创建一个存储函数，以便在使用冒号 col1 拆分列 col1 之后通过使用条件聚合动态地转换结果，例如

CREATE OR REPLACE FUNCTION Get_Splitted_Columns RETURN SYS_REFCURSOR IS
  v_recordset SYS_REFCURSOR;
  v_sql       VARCHAR2(32767);
  v_cols      VARCHAR2(32767);
BEGIN
  SELECT LISTAGG('MAX( CASE WHEN rn2 =  '''||level||''' THEN col1 END ) AS "L_'||level||'"' , ',' )
                 WITHIN GROUP ( ORDER BY level )
    INTO v_cols
    FROM dual
   CONNECT BY level <= ( SELECT MAX(REGEXP_COUNT(col1,':',INSTR(col1,col2))) FROM tab );
                            
  v_sql :='WITH t AS
          (
           SELECT REGEXP_SUBSTR(col1,''[^:]+'',1,level) AS col1, col2, level AS rn,
                  DENSE_RANK() OVER (ORDER BY col1) AS dr              
             FROM tab
          CONNECT BY level <= REGEXP_COUNT(col1,'':'')+1
              AND PRIOR SYS_GUID() IS NOT NULL
              AND PRIOR col1 = col1
          ), t2 AS
          (
           SELECT t.*,MAX(CASE WHEN col1=col2 THEN rn END) OVER (PARTITION BY dr) AS max_rn
             FROM t
          ), t3 AS
          (
           SELECT dr, max_rn, rn, rn - max_rn AS rn2, col1, col2 
             FROM t2 tt
            WHERE rn > ( SELECT MAX(max_rn) FROM t2 WHERE dr = tt.dr )
          )
          SELECT '||v_cols||'
            FROM t3
           GROUP BY dr';

  OPEN v_recordset FOR v_sql;
  RETURN v_recordset;
END;
/

然后将函数调用为

VAR rc REFCURSOR
EXEC :rc := Get_Splitted_Columns;
PRINT rc

来自SQL开发者的命令行

Demo for Generated SQL by Function

Answer 3

好吧，如果没有超过 L5 的东西，并且您总是有三个字母，只需使用第一个匹配项的偏移量对其进行硬编码。

WITH tab AS (
  SELECT 'ABC:EFG:MNO:XYZ'         AS Col1, 'MNO' AS Col2 FROM DUAL
  UNION
  SELECT 'PQR:NOM:XYN:SDF:RST:EDF' AS Col1, 'NOM' AS Col2 FROM DUAL
)

SELECT tab.Col1,
       tab.Col2,
       SUBSTR(tab.Col1, NULLIF(INSTR(tab.Col1, tab.Col2), 0) +  4, 3) AS L1,
       SUBSTR(tab.Col1, NULLIF(INSTR(tab.Col1, tab.Col2), 0) +  8, 3) AS L2,
       SUBSTR(tab.Col1, NULLIF(INSTR(tab.Col1, tab.Col2), 0) + 12, 3) AS L3,
       SUBSTR(tab.Col1, NULLIF(INSTR(tab.Col1, tab.Col2), 0) + 16, 3) AS L4,
       SUBSTR(tab.Col1, NULLIF(INSTR(tab.Col1, tab.Col2), 0) + 20, 3) AS L5
FROM   tab;

您找到与 INSTR 的第一个匹配项，并使用从匹配项 + 条目 * 4 开始的 3 个字符的子字符串。NULLIF 表示没有匹配项。

Oracle 中的子字符串查询 SQL

Substring Query in Oracle SQL

sql

oracle

split

pivot

oracle11g