I have an SQL table with 11000 keywords in it.
I want a query that can find fields which contain a certain letter.
So, if I include "a" and "b" the query will select all fields which contain the letter "a" and the letter "b" somewhere in the field.
This question is related to
sql
All the answers given using LIKE
are totally valid, but as all of them noted will be slow. So if you have a lot of queries and not too many changes in the list of keywords, it pays to build a structure that allows for faster querying.
Here are some ideas:
If all you are looking for is the letters a-z and you don't care about uppercase/lowercase, you can add columns containsA .. containsZ and prefill those columns:
UPDATE table
SET containsA = 'X'
WHERE UPPER(your_field) Like '%A%';
(and so on for all the columns).
Then index the contains.. columns and your query would be
SELECT
FROM your_table
WHERE containsA = 'X'
AND containsB = 'X'
This may be normalized in an "index table" iTable with the columns your_table_key
, letter, index the letter-column and your query becomes something like
SELECT
FROM your_table
WHERE <key> in (select a.key
From iTable a join iTable b and a.key = b.key
Where a.letter = 'a'
AND b.letter = 'b');
All of these require some preprocessing (maybe in a trigger or so), but the queries should be a lot faster.
There's most likely a more elegant way, but this does find every record with a letter in it, both upper or lower case:
select * from your_table
where UPPER(your_field) like '%A%'
or UPPER(your_field) like '%B%'
or UPPER(your_field) like '%C%'
or UPPER(your_field) like '%D%'
or UPPER(your_field) like '%E%'
or UPPER(your_field) like '%F%'
or UPPER(your_field) like '%G%'
or UPPER(your_field) like '%H%'
or UPPER(your_field) like '%I%'
or UPPER(your_field) like '%J%'
or UPPER(your_field) like '%K%'
or UPPER(your_field) like '%L%'
or UPPER(your_field) like '%M%'
or UPPER(your_field) like '%N%'
or UPPER(your_field) like '%O%'
or UPPER(your_field) like '%P%'
or UPPER(your_field) like '%Q%'
or UPPER(your_field) like '%R%'
or UPPER(your_field) like '%S%'
or UPPER(your_field) like '%T%'
or UPPER(your_field) like '%U%'
or UPPER(your_field) like '%V%'
or UPPER(your_field) like '%W%'
or UPPER(your_field) like '%X%'
or UPPER(your_field) like '%Y%'
or UPPER(your_field) like '%Z%'
I'd use wildcard searching.
where <field> like '%[ab]%'
It isn't regex, but it does a good job.
You can also do variants like <field> like 'sim[oa]ns'
-- which will match simons, and simans...
Depnding on your collation you may or may not have to include case data, like '%[aAbB]%'
As mentioned elsewhere be prepared for a wait since indexes are out of the question when you're doing contains searching.
select *
from table
where keyword like '%a%'
and keyword like '%b%'
ps This will be super slow. You may want to investigate full text indexing solutions.
I'll assume you meant more or less what you said, and you want to find keywords in your table that "contain the letter 'a' and the letter 'b'." Some of the solutions here give the answer to a different question.
To get keywords that contain both the letters 'a' and 'b' in them (as opposed to those that contain either letter), you can use 'ab' as the in the query below:
select
keyword
from myTable
where not exists (
select Nums26.i from Nums26
where Nums26.i <= len(<matchsetstring>) -- or your dialect's equivalent for LEN()
and keyword not like '%'+substring(<matchsetstring>,Nums26.i,1)+'%' -- adapt SUBSTRING to your dialect
);
The table named "Nums26" should contain a column "i" (indexed for efficiency) that contains each of the values 1 through 26 (or more if you might try to match more than letters). See below. Advice given by others applies with regard to upper/lower case. If your collation is case-sensitive, however, you can't simply specify 'aAbB' here as your , because that would request keywords that contain each of the four characters a, A, b, and B. You might use UPPER and match 'AB', perhaps.
create table nums26 (
i int primary key
);
insert into nums26 values (1);
insert into nums26 select 1+i from nums26;
insert into nums26 select 2+i from nums26;
insert into nums26 select 4+i from nums26;
insert into nums26 select 8+i from nums26;
insert into nums26 select 16+i from nums26;
select * from your_table where your_field like '%a%b%'
and be prepared to wait a while...
Edit: note that this pattern looks for an 'a' followed by a 'b' (possibly with other "stuff" in between) -- rereading your question, that may not be what you wanted...
where somefield like '%a%' or somefield like '%b%'
try this
Select * From Table
Where field like '%' + ltrValue1 + '%'
And field like '%' + ltrValue2 + '%'
... etc.
and be prepared for a table scan as this functionality cannot use any existing indices
You can use a cursor and temp table approach so you aren't doing full table scan each time. What this would be doing is populating the temp table with all of your keywords, and then with each string in the @letters XML
, it would remove any records from the temp table. At the end, you only have records in your temp table that have each of your desired strings in it.
declare @letters xml
SET @letters = '<letters>
<letter>a</letter>
<letter>b</letter>
</letters>'
-- SELECTING LETTERS FROM THE XML
SELECT Letters.l.value('.', 'nvarchar(50)') AS letter
FROM @letters.nodes('/letters/letter') AS Letters(l)
-- CREATE A TEMP TABLE WE CAN DELETE FROM IF A RECORD DOESN'T HAVE THE LETTER
CREATE TABLE #TempResults (keywordID int not null, keyWord nvarchar(50) not null)
INSERT INTO #TempResults (keywordID, keyWord)
SELECT employeeID, firstName FROM Employee
-- CREATE A CURSOR, SO WE CAN LOOP THROUGH OUR LETTERS AND REMOVE KEYWORDS THAT DON'T MATCH
DECLARE Cursor_Letters CURSOR READ_ONLY
FOR
SELECT Letters.l.value('.', 'nvarchar(50)') AS letter
FROM @letters.nodes('/letters/letter') AS Letters(l)
DECLARE @letter varchar(50)
OPEN Cursor_Letters
FETCH NEXT FROM Cursor_Letters INTO @letter
WHILE (@@fetch_status <> -1)
BEGIN
IF (@@fetch_status <> -2)
BEGIN
DELETE FROM #TempResults
WHERE keywordID NOT IN
(SELECT keywordID FROM #TempResults WHERE keyWord LIKE '%' + @letter + '%')
END
FETCH NEXT FROM Cursor_Letters INTO @letter
END
CLOSE Cursor_Letters
DEALLOCATE Cursor_Letters
SELECT * FROM #TempResults
DROP Table #TempResults
GO
Source: Stackoverflow.com