SQL Server replace, remove all after certain character

Question

My data looks like

ID    MyText
1     some text; some more text
2     text again; even more text

How can I update MyText to drop everything after the semi-colon and including the semi colon, so I'm left with the following:

ID    MyText
1     some text
2     text again

I've looked at SQL Server Replace, but can't think of a viable way of checking for the ";"

Solution 1

Use LEFT combined with CHARINDEX:

UPDATE MyTable
SET MyText = LEFT(MyText, CHARINDEX(';', MyText) - 1)
WHERE CHARINDEX(';', MyText) > 0

Note that the WHERE clause skips updating rows in which there is no semicolon.

Here is some code to verify the SQL above works:

declare @MyTable table ([id] int primary key clustered, MyText varchar(100))
insert into @MyTable ([id], MyText)
select 1, 'some text; some more text'
union all select 2, 'text again; even more text'
union all select 3, 'text without a semicolon'
union all select 4, null -- test NULLs
union all select 5, '' -- test empty string
union all select 6, 'test 3 semicolons; second part; third part;'
union all select 7, ';' -- test semicolon by itself    

UPDATE @MyTable
SET MyText = LEFT(MyText, CHARINDEX(';', MyText) - 1)
WHERE CHARINDEX(';', MyText) > 0

select * from @MyTable

I get the following results:

id MyText
-- -------------------------
1  some text
2  text again
3  text without a semicolon
4  NULL
5        (empty string)
6  test 3 semicolons
7        (empty string)

Solution 2

For the times when some fields have a ";" and some do not you can also add a semi-colon to the field and use the same method described.

SET MyText = LEFT(MyText+';', CHARINDEX(';',MyText+';')-1)

Solution 3

Could use CASE WHEN to leave those with no ';' alone.

    SELECT
    CASE WHEN CHARINDEX(';', MyText) > 0 THEN
    LEFT(MyText, CHARINDEX(';', MyText)-1) ELSE
    MyText END
    FROM MyTable

Solution 4

UPDATE MyTable
   SET MyText = SUBSTRING(MyText, 1, CHARINDEX(';', MyText) - 1)
 WHERE CHARINDEX(';', MyText) > 0 

Solution 5

Use CHARINDEX to find the ";". Then use SUBSTRING to just return the part before the ";".

View more solutions

Comments

[Mike]

What if MyText does not contain the ';' character? In that case, wouldn't you be using a negative 1 as the second parameter in left(). In that case, on my box, I get an error of "Invalid length parameter passed to the substring function."

[Etienne Dupuis]

@Mike That is exactly my problem atm - See next Answer from @najmeddine!!

[Tim]

@Mike Rashlien's response below resolves this: LEFT(MyText+';', CHARINDEX(';',MyText+';')-1)

[Paul Williams]

Note that the WHERE clause above skips rows that do not have a semicolon. The OP wanted to remove the semicolon and any text after it. If there is no semicolon, there is nothing to remove, so the UPDATE statement will skip that row.

[Paul Williams]

Updated this answer with some test cases to show how it works with missing semicolons, NULLs, empty strings, and additional semicolons.