CODEWITHSUNDEEP
mysql
Christopher
Christopher

Reputation:

Search for text between delimiters in MySQL

I am trying to extract a certain part of a column that is between delimiters.

e.g. find foo in the following

test 'esf :foo: bar

So in the above I'd want to return foo, but all the regexp functions only return true|false, is there a way to do this in MySQL

Upvotes: 13

Views: 25964

Answers (11)

mprot
mprot

Reputation: 41

This one looks elegant to me. Strip all after n-th separator, rotate string, strip everything after 1. separator, rotate back. 

select
  reverse(
    substring_index(
      reverse(substring_index(str,separator,substrindex)),
      separator,
      1)
  );

For example: 

select
  reverse(
    substring_index(
      reverse(substring_index('www.mysql.com','.',2)),
      '.',
      1
    )
  );

Upvotes: 0

brewmanz
brewmanz

Reputation: 1221

This is what I am extracting from (mainly colon ':' as delimiter but some exceptions), as column theline255 in table loaddata255:

23856.409:0023:trace:message:SPY_EnterMessage (0x2003a) L"{#32769}"      [0081] WM_NCCREATE sent from self wp=00000000 lp=0023f0b0

This is the MySql code (It quickly did what I want, and is straight forward):

select 
time('2000-01-01 00:00:00' + interval substring_index(theline255, '.', 1) second) as hhmmss
, substring_index(substring_index(theline255, ':', 1), '.', -1) as logMilli
, substring_index(substring_index(theline255, ':', 2), ':', -1) as logTid
, substring_index(substring_index(theline255, ':', 3), ':', -1) as logType
, substring_index(substring_index(theline255, ':', 4), ':', -1) as logArea
, substring_index(substring_index(theline255, ' ', 1), ':', -1) as logFunction
, substring(theline255, length(substring_index(theline255, ' ', 1)) + 2) as logText
from loaddata255

and this is the result:

# LogTime, LogTimeMilli, LogTid, LogType, LogArea, LogFunction, LogText
'06:37:36', '409', '0023', 'trace', 'message', 'SPY_EnterMessage', '(0x2003a) L\"{#32769}\"      [0081] WM_NCCREATE sent from self wp=00000000 lp=0023f0b0'

Upvotes: 0

hrushikesh
hrushikesh

Reputation: 317

mid(col, 
    locate('?m=',col) + char_length('?m='), 
    locate('&o=',col) - locate('?m=',col) - char_length('?m=') 
)

A bit compact form by replacing char_length(.) with the number 3 

mid(col, locate('?m=',col) + 3, locate('&o=',col) - locate('?m=',col) - 3)

the patterns I have used are '?m=' and '&o'.

Upvotes: 2

Danny Z
Danny Z

Reputation: 123

This should work if the two delimiters only appear twice in your column. I am doing something similar...

substring_index(substring_index(column,':',-2),':',1)

Upvotes: 7

chris2k
chris2k

Reputation: 169

you can use the substring / locate function in 1 command

here is a mice tutorial:

http://infofreund.de/mysql-select-substring-2-different-delimiters/

The command as describes their should look for u:

**SELECT substr(text,Locate(' :', text )+2,Locate(': ', text )-(Locate(' :', text )+2)) FROM testtable**

where text is the textfield which contains "test 'esf :foo: bar"

So foo can be fooooo or fo - the length doesnt matter :).

Upvotes: -2

micahwittman
micahwittman

Reputation: 12476

If you know the position you want to extract from as opposed to what the data itself is:

$colNumber = 2; //2nd position
$sql = "REPLACE(SUBSTRING(SUBSTRING_INDEX(fooField, ':', $colNumber),
                             LENGTH(SUBSTRING_INDEX(fooField, 
                                                    ':', 
                                                    $colNumber - 1)) + 1)";

Upvotes: 0

Pete Karl II
Pete Karl II

Reputation: 4300

Here ya go, bud:

SELECT 
  SUBSTR(column, 
    LOCATE(':',column)+1, 
      (CHAR_LENGTH(column) - LOCATE(':',REVERSE(column)) - LOCATE(':',column))) 
FROM table

Yea, no clue why you're doing this, but this will do the trick.

By performing a LOCATE, we can find the first ':'. To find the last ':', there's no reverse LOCATE, so we have to do it manually by performing a LOCATE(':', REVERSE(column)).

With the index of the first ':', the number of chars from the last ':' to the end of the string, and the CHAR_LENGTH (don't use LENGTH() for this), we can use a little math to discover the length of the string between the two instances of ':'.

This way we can peform a SUBSTR and dynamically pluck out the characters between the two ':'.

Again, it's gross, but to each his own.

Upvotes: 23

Adam Bellaire
Adam Bellaire

Reputation: 110499

With only one set of delimeters, the following should work:

SUBSTR(
    SUBSTR(fooField,LOCATE(':',fooField)+1),
    1,
    LOCATE(':',SUBSTR(fooField,LOCATE(':',fooField)+1))-1
 )

Upvotes: 2

Vinko Vrsalovic
Vinko Vrsalovic

Reputation: 340321

select mid(col from locate(':',col) + 1 for 
locate(':',col,locate(':',col)+1)-locate(':',col) - 1 ) 
from table where col rlike ':.*:';

Upvotes: 0

Adam Bellaire
Adam Bellaire

Reputation: 110499

I don't know if you have this kind of authority, but if you have to do queries like this it might be time to renormalize your tables, and have these values in a lookup table.

Upvotes: 2

Mark Biek
Mark Biek

Reputation: 150799

A combination of LOCATE and MID would probably do the trick.

If the value "test 'esf :foo: bar" was in the field fooField:

MID( fooField, LOCATE('foo', fooField), 3);

Upvotes: 3

Related Questions