CODEWITHSUNDEEP
sqlregexoracle-database
Brian Zhang
Brian Zhang

Reputation: 3

How to select the list of words containing a particular substring as part of a SQL query (oracle)?

I'm trying to return the list of "words" (separated by spaces) containing a certain substring within a string as part of an Oracle Sql query. Would like to return the result as a comma separated list. Separate rows for each match would also work.

Example String in [text_col] field:

some words 123-asdf-789A and also this one 456-asdf-555A more words etc.

Desired result: 123-asdf-789A, 456-asdf-555A

This is what I have so far but it only returns the first result and the fact that it's two separate regular expressions makes it difficult to concatenate all matches as I would like to do.

CONCAT(REGEXP_SUBSTR(text_col, ''(([^[:space:]]+)\asdf)'', 1, 1, ''i'', 1), 
REGEXP_SUBSTR(text_col, ''\asdf([^[:space:]]+)'', 1, 1, ''i'', 1))

Upvotes: 0

Views: 219

Answers (2)

MT0
MT0

Reputation: 167981

Use a recursive sub-query factoring clause and iterate through all the matches concatenating the string as you go:

Oracle Setup:

CREATE TABLE test_data ( value ) AS
SELECT 'some words 123-asdf-789A and also this one 456-asdf-555A more words etc.' FROM DUAL UNION ALL
SELECT 'some words without the expected sub-string' FROM DUAL UNION ALL
SELECT 'asdf asdf-123 456-asdf 78-asdf-90' FROM DUAL

Query:

WITH matches ( value, idx, cnt, match ) AS (
  SELECT value,
         0,
         REGEXP_COUNT( value, '\S*asdf\S*' ),
         CAST( NULL AS VARCHAR2(4000) )
  FROM   test_data
UNION ALL
  SELECT value,
         idx + 1,
         cnt,
         CASE idx WHEN 0 THEN '' ELSE match || ' ' END
           || REGEXP_SUBSTR( value, '\S*asdf\S*', 1, idx + 1 )
  FROM   matches
  WHERE  idx < cnt
)
SELECT value, match
FROM   matches
WHERE  idx = cnt;

Output:

VALUE                                                                    | MATCH                            
:----------------------------------------------------------------------- | :--------------------------------
some words without the expected sub-string                               | null                             
some words 123-asdf-789A and also this one 456-asdf-555A more words etc. | 123-asdf-789A 456-asdf-555A      
asdf asdf-123 456-asdf 78-asdf-90                                        | asdf asdf-123 456-asdf 78-asdf-90

db<>fiddle here

Upvotes: 0

Barbaros &#214;zhan
Barbaros &#214;zhan

Reputation: 65278

You can use some regexp functions together as :

with tab(str) as
(
 select 'some words 123-asdf-789A and also this one 456-asdf-555A more words etc' from dual
), t as 
(
select regexp_substr(str,'[^[:space:]]+',1,level) as str, level as lvl
  from tab
connect by level <= regexp_count(str,'[:space:]')  
)
select listagg(str,',') within group (order by lvl) as "Result"
  from t
 where regexp_like(str,'-');

Result
---------------------------------
123-asdf-789A,456-asdf-555A

Demo

first split by spaces (through [:space:] posix) and take the ones containing dash characters, and finally concatenate by listagg() function

Upvotes: 1

Related Questions