I have a varchar(50) sql server column with data like this:
RawData
----------------------------
Washington 40 New Orleans 32
Detroit 27 St. Louis 23
Here is a very ugly implementation of PATINDEX()
which returns the data in the multiple columns:
SELECT rtrim(substring(RawData, 1, PATINDEX('%[0-9]%', RawData) -1)) WinningTeam,
ltrim(rtrim(substring(RawData, PATINDEX('%[0-9]%', RawData), 2))) WinningTeamScore,
reverse(ltrim(rtrim(substring(reverse(SUBSTRING(RawData, 1, LEN(rawdata) - (PATINDEX('%[aA-zZ]%', REVERSE(rawData)) - PATINDEX('%[0-9]%', reverse(RawData))))), 1, PATINDEX('%[0-9]%', SUBSTRING(RawData, 1, LEN(rawdata) - (PATINDEX('%[aA-zZ]%', REVERSE(rawData)) - PATINDEX('%[0-9]%', reverse(RawData))))))))) LosingTeam,
substring(reverse(RawData), PATINDEX('%[0-9]%', reverse(RawData)), (PATINDEX('%[aA-zZ]%', REVERSE(rawData)) - PATINDEX('%[0-9]%', reverse(RawData)))) LosingTeamScore
from yourtable
See SQL Fiddle with Demo
You can use the PATINDEX function instead of CHARINDEX, here lies the documentation ;)
Maybe a bit complicated but it works for what you need:
declare @test table(mytext varchar(50))
insert @test values('Washington 40 New Orleans 32')
insert @test values('Detroit 27 St. Louis 23')
select
WinningTeam=SubString(mytext, 0,PatIndex('%[0-9.-]%', mytext)),
WinningTeamScore=Left(SubString(mytext, PatIndex('%[0-9.-]%', mytext), 50),PatIndex('%[^0-9.-]%', SubString(mytext, PatIndex('%[0-9.-]%', mytext), 50) + 'X')-1),
LosingTeam=SubString(mytext, PatIndex('%[0-9.-]%', mytext)+3,PatIndex('%[0-9.-]%', mytext)),
LosingTeamScore=reverse(Left(SubString(reverse(mytext), PatIndex('%[0-9.-]%', reverse(mytext)), 50),PatIndex('%[^0-9.-]%', SubString(reverse(mytext), PatIndex('%[0-9.-]%', reverse(mytext)), 50) + 'X')-1))
from @test
The query above works for scores under 100 points but you can modify it to deal with any number.
my query for selecting multi-didget numbers out of a url (ignore rows without a number and rows where the number is after the '?'
select
URL,
substring(URL,PATINDEX ('%[0-9][0-9][0-9][0-9][0-9][0-9][0-9]%',URL),7) as id
from data
where
PATINDEX ('%[0-9][0-9][0-9][0-9][0-9][0-9][0-9]%',URL)>0
and PATINDEX ('%[0-9][0-9][0-9][0-9][0-9][0-9][0-9]%',URL)<charindex ('?',URL)
Is there a way to identify the position of the first number in a string?
Yes
SELECT PATINDEX('%[0-9]%','Washington 40 New Orleans 32')
PATINDEX
returns 0 if the pattern can't be found or the 1
based index of the beginning of the match otherwise.