How to parse data using REGEXP_SUBSTR?

问题

I have a data set like this (see below) and I try to extract digits which are in form {variable_number_of_digits}{hyphen}{only_one_digit}:

with mcte as (
select 'ILLD/ELKJS/00000000/ELKJS/FHSH' as addr from dual
union all 
select 'ILLD/EFECTE/0116988-7-002/ADFA/ADFG' as addr from dual
union all
select 'IIODK/1573230-0/2216755-7/' as addr  from dual
union all
select 'IIODK/1573230-0/2216755-700/WRITE' as addr from dual
)
select  addr, 
        REGEXP_SUBSTR(addr,'(\/)([0-9-]+)',1,1,NULL,2) AS num1,
        REGEXP_SUBSTR(addr,'(\/)([^\/]+\/)([0-9\-]+)',1,1,NULL,3) num2
from mcte
;

I'm not getting a correct results set which should be the following

+-------------------------------------+-----------+-----------+
|                ADDR                 |   NUM1    |   NUM2    |
+-------------------------------------+-----------+-----------+
| ILLD/ELKJS/00000000/ELKJS/FHSH      | NULL      | NULL      |
| ILLD/EFECTE/0116988-7-002/ADFA/ADFG | NULL      | NULL      |
| IIODK/1573230-0/2216755-7/          | 1573230-0 | 2216755-7 |
| IIODK/1573230-0/2216755-700/WRITE   | 1573230-0 | NULL      |
+-------------------------------------+-----------+-----------+

How to achieve this?

回答1:

If you want to get the results from the second and third / delimited groups then:

with mcte ( addr ) as (
  select 'ILLD/ELKJS/00000000/ELKJS/FHSH'      from dual union all 
  select 'ILLD/EFECTE/0116988-7-002/ADFA/ADFG' from dual union all
  select 'IIODK/1573230-0/2216755-7/'          from dual union all
  select 'IIODK/1573230-0/2216755-700/WRITE'   from dual union all
  select 'IIODK/TEST/1573230-0/2216755-700/WRITE'   from dual
)
select  addr, 
        REGEXP_SUBSTR(addr,'^[^/]*/(\d+-\d)/',1,1,NULL,1) AS num1,
        REGEXP_SUBSTR(addr,'^[^/]*/[^/]*/(\d+-\d)/',1,1,NULL,1) num2
from mcte;

Output:

ADDR                                   NUM1                NUM2
-------------------------------------- ------------------- -------------------
ILLD/ELKJS/00000000/ELKJS/FHSH
ILLD/EFECTE/0116988-7-002/ADFA/ADFG
IIODK/1573230-0/2216755-7/             1573230-0           2216755-7
IIODK/1573230-0/2216755-700/WRITE      1573230-0
IIODK/TEST/1573230-0/2216755-700/WRITE                     1573230-0

Update:

If you just want the first and second pattern that match and do not care where they are in the string then:

with mcte ( addr ) as (
  select 'ILLD/ELKJS/00000000/ELKJS/FHSH'         from dual union all 
  select 'ILLD/EFECTE/0116988-7-002/ADFA/ADFG'    from dual union all
  select 'IIODK/1573230-0/2216755-7/'             from dual union all
  select 'IIODK/1573230-0/2216755-700/WRITE'      from dual union all
  select 'IIODK/TEST/1573230-0/2216755-700/WRITE' from dual union all
  select '1234567-8'                              from dual union all
  select '1234567-8/9876543-2'                    from dual union all
  select '1234567-8/TEST/9876543-2'               from dual
)
select  addr, 
        REGEXP_SUBSTR(addr,'(^|/)(\d+-\d)(/|$)',1,1,NULL,2) AS num1,
        REGEXP_SUBSTR(addr,'(^|/)\d+-\d(/.+?)?/(\d+-\d)(/|$)',1,1,NULL,3) num2
from mcte;

Outputs:

ADDR                                   NUM1                NUM2

-------------------------------------- ------------------- ------------------
ILLD/ELKJS/00000000/ELKJS/FHSH
ILLD/EFECTE/0116988-7-002/ADFA/ADFG
IIODK/1573230-0/2216755-7/             1573230-0           2216755-7
IIODK/1573230-0/2216755-700/WRITE      1573230-0
IIODK/TEST/1573230-0/2216755-700/WRITE 1573230-0           
1234567-8                              1234567-8
1234567-8/9876543-2                    1234567-8           9876543-2
1234567-8/TEST/9876543-2               1234567-8           9876543-2

回答2:

I try to extract digits which are in form {variable_number_of_digits}{hyphen}{only_one_digit}

To match numbers in this format you should do something like this.

Regex: \/\d+-\d

Regex101 Demo

回答3:

Combining the delimiter split query with REGEXP_LIKE and pivot-ing the result you get this query working for up to 6 numbers. You will need to update the cols subquery and teh pivot list to be able to process more numbers per record. (Unfortunately this can't be done general in a static SQL).

with mcte as (
  select 1 id, 'ILLD/ELKJS/00000000/ELKJS/FHSH' as addr from dual
  union all 
  select 2 id, 'ILLD/EFECTE/0116988-7-002/ADFA/ADFG' as addr from dual
  union all
  select 3 id, 'IIODK/1573230-0/2216755-7/' as addr  from dual
  union all
  select 4 id, '1-1/1573230-0/2216755-700/676-7' as addr from dual
),
cols as (select  rownum colnum from dual connect by level < 6 /* (max) number of columns */),
mcte2 as (select id, cols.colnum, (regexp_substr(addr,'[^/]+', 1, cols.colnum)) addr 
              from mcte, cols where regexp_substr(addr, '[^/]+', 1, cols.colnum) is not null),
mcte3 as (              
select ID, 
ROW_NUMBER() over (partition by ID order by COLNUM) as col_no, ADDR from mcte2
where REGEXP_like(addr, '^[0-9]+-[0-9]$')
)
select * from mcte3
PIVOT (max(addr)   for (col_no) in 
     (1 as "NUM1",
      2 as "NUM2",
      3 as "NUM3",
      4 as "NUM4",
      5 as "NUM5",
      6 as "NUM6"))
order by id;

this gives a result

        ID NUM1       NUM2       NUM3       NUM4       NUM5       NUM6     
---------- ---------- ---------- ---------- ---------- ---------- ----------
         3 1573230-0  2216755-7                                              
         4 1-1        1573230-0  676-7       

来源：https://stackoverflow.com/questions/36292350/how-to-parse-data-using-regexp-substr