在 Oracle SQL 中按不同值分组和排序

Question

我有 table 这样的：

id	full_name
1	John Smith
2	Smith John
3	Jim Jonson
4	JimJonson

我想得到这样的东西：

id	full_name
1	John Smith
3	Jim Jonson

所以，我需要SELECT DISTINCT full_name FROM table，这样

John Smith和Smith John是一回事，也是
Jim Jonson 和 JimJonson

希望我解释得很好。你能帮帮我吗？

Answer 1

循序渐进。阅读代码中的注释。

SQL> with test (id, full_name) as
  2    -- sample data
  3    (select 1, 'John Smith' from dual union all
  4     select 2, 'Smith John' from dual union all
  5     select 3, 'Jim Jonson' from dual union all
  6     select 4, 'JimJonson'  from dual
  7    ),

  8  temp as
  9    -- split full name to rows
 10    (select id,
 11       regexp_substr(full_name, '[^ ]+', 1, column_value) val,
 12      column_value cv
 13     from test cross join
 14       table(cast(multiset(select level from dual
 15                           connect by level <= regexp_count(full_name, ' ') + 1
 16                          ) as sys.odcinumberlist))
 17    ),
 18  temp2 as
 19    -- aggregate full name with no space between "words"
 20    (select id,
 21       listagg(val, '') within group (order by val) full_name
 22     from temp
 23     group by id
 24    ),
 25  temp3 as
 26    -- fetch only distinct values
 27    (select min(b.id) id,
 28            b.full_name
 29     from temp2 b
 30     group by b.full_name
 31    )

 32  -- finally, join TEMP3 and sample data
 33  select b.id,
 34         a.full_name
 35  from test a join temp3 b on b.id = a.id
 36  order by a.id;

        ID FULL_NAME
---------- ----------------------------------------
         1 John Smith
         3 Jim Jonson

SQL>

Answer 2

您可以将 full_name 值按姓名和姓氏的首字母大写拆分，并按字母顺序排序，然后使用 LISTAGG() 函数合并，并在以下位置应用 MIN() 聚合最后一步如

WITH t(id,full_name) AS
(
 SELECT 1, 'John Smith' FROM dual UNION ALL 
 SELECT 2, 'Smith John' FROM dual UNION ALL
 SELECT 3, 'Jim Jonson' FROM dual UNION ALL
 SELECT 4, 'JimJonson'  FROM dual 
), t2 AS
(
  SELECT id,  
         TRIM(SUBSTR(full_name,column_value,LEAD(column_value,1,LENGTH(full_name)) OVER (PARTITION BY id ORDER BY id,column_value)-1)) AS names
    FROM t,
         TABLE(CAST(MULTISET(SELECT REGEXP_INSTR(full_name,'[A-Z]+',1,level)
                               FROM dual 
                            CONNECT BY level <= REGEXP_COUNT(full_name,'[A-Z]')) AS sys.odcivarchar2list ))                        
), t3 AS
(
 SELECT id, LISTAGG(names,' ') WITHIN GROUP (ORDER BY id,names) AS full_name
   FROM t2
  GROUP BY id                           
)
SELECT MIN(id) AS min_id, full_name
  FROM t3 
 GROUP BY full_name
 ORDER BY min_id

Demo

在 Oracle SQL 中按不同值分组和排序

Grouping and Sorting By distinct values in Oracle's SQL

sql

oracle

group-by

aggregate-functions