Can CONTAINSTABLE be made to match unconditionally?

Say that users searching a Books database are able to search by Author,
Title, and Publisher using three separate text boxes. They can choose to
leave any or all text boxes empty if they do not care to filter by that
particular item.
Say I have a query that looks similar to this:
b.* FROM Books b
INNER JOIN CONTAINSTABLE( Books, Author, @.author) authorRank ON b.BookId =
authorRank.[KEY]
INNER JOIN CONTAINSTABLE( Books, Title, @.title) titleRank ON ( b.BookId =
titleRank.[KEY] )
INNER JOIN CONTAINSTABLE( Books, Publisher, @.publisher) publisherRank ON (
b.BookId = publisherRank.[KEY] )
This works great if the user chooses to enter something in each of the three
text boxes. However, CONTAINSTABLE does not accept an empty string for the
search condition (nor does it accept *, %, or other wildcards except when
searching for a prefix). How can I get the query to unconditionally match
title and publisher if they leave title and publisher blank but type
something for author?
If there is no way to allow any title and publisher to be returned when the
user leaves those text boxes blank, I will need to write a query for if they
type all three text boxes, a query for if they type in none of the text
boxes, a query if they type in just the publisher but not the title or
author, a query... etc. Not a good solution.
How can I achieve the results I want without writing multiple queries?
Thank you.Hi, Greg
Try something like this (untested):
SELECT * FROM Books
WHERE (@.author IS NULL OR BookId IN (
SELECT KEY FROM CONTAINSTABLE(Books, Author, @.author)
)) AND (@.title IS NULL OR BookId IN (
SELECT KEY FROM CONTAINSTABLE(Books, Title, @.title)
)) AND (@.publisher IS NULL OR BookId IN (
SELECT KEY FROM CONTAINSTABLE(Books, Publisher, @.publisher)
))
Razvan|||That doesn't let me use any of the ranking information provided by
CONTAINSTABLE. It may be workable. I may have found some sort of
sp_configure setting that will change the behavior of noise words. I'm not
sure about it yet.
"Razvan Socol" wrote:

> Hi, Greg
> Try something like this (untested):
> SELECT * FROM Books
> WHERE (@.author IS NULL OR BookId IN (
> SELECT KEY FROM CONTAINSTABLE(Books, Author, @.author)
> )) AND (@.title IS NULL OR BookId IN (
> SELECT KEY FROM CONTAINSTABLE(Books, Title, @.title)
> )) AND (@.publisher IS NULL OR BookId IN (
> SELECT KEY FROM CONTAINSTABLE(Books, Publisher, @.publisher)
> ))
> Razvan
>|||> That doesn't let me use any of the ranking information provided by
> CONTAINSTABLE.
In this case, you might want to use something like this (also
untested):
SELECT b.*, x.RANK, y.RANK, z.RANK FROM Books b
LEFT JOIN CONTAINSTABLE(Books, Author, @.author) x ON b.BookID=x.[KEY]
LEFT JOIN CONTAINSTABLE(Books, Title, @.title) y ON b.BookID=y.[KEY]
LEFT JOIN CONTAINSTABLE(Books, Publisher, @.publisher) z ON
b.BookID=z.[KEY]
WHERE (@.author IS NULL OR x.[KEY] IS NOT NULL)
AND (@.title IS NULL OR y.[KEY] IS NOT NULL)
AND (@.publisher IS NULL OR z.[KEY] IS NOT NULL)
Razvan|||If @.author is NULL (or blank), then CONTAINSTABLE will throw an error, which
is pretty much the entire problem here.
"Razvan Socol" wrote:

> In this case, you might want to use something like this (also
> untested):
> SELECT b.*, x.RANK, y.RANK, z.RANK FROM Books b
> LEFT JOIN CONTAINSTABLE(Books, Author, @.author) x ON b.BookID=x.[KEY]
> LEFT JOIN CONTAINSTABLE(Books, Title, @.title) y ON b.BookID=y.[KEY]
> LEFT JOIN CONTAINSTABLE(Books, Publisher, @.publisher) z ON
> b.BookID=z.[KEY]
> WHERE (@.author IS NULL OR x.[KEY] IS NOT NULL)
> AND (@.title IS NULL OR y.[KEY] IS NOT NULL)
> AND (@.publisher IS NULL OR z.[KEY] IS NOT NULL)
> Razvan
>|||> If @.author is NULL (or blank), then CONTAINSTABLE will throw an error [...]
Aha! I told you it was untested... :) I assumed that CONTAINSTABLE will
return an empty resultset when given a NULL search condition.
Obviously, I was wrong.
In this case, we can use a non-existent word instead of NULL, like
this:
SET @.title=ISNULL(@.title,'NotSpecified')
SET @.author=ISNULL(@.author,'NotSpecified')
SET @.publisher=ISNULL(@.publisher,'NotSpecifi
ed')
SELECT b.*, x.RANK, y.RANK, z.RANK FROM Books b
LEFT JOIN CONTAINSTABLE(Books, Author, @.author) x ON b.BookId=x.[KEY]
LEFT JOIN CONTAINSTABLE(Books, Title, @.title) y ON b.BookId=y.[KEY]
LEFT JOIN CONTAINSTABLE(Books, Publisher, @.publisher) z ON
b.BookId=z.[KEY]
WHERE (@.author='NotSpecified' OR x.[KEY] IS NOT NULL)
AND (@.title='NotSpecified' OR y.[KEY] IS NOT NULL)
AND (@.publisher='NotSpecified' OR z.[KEY] IS NOT NULL)
If you think the word "NotSpecified" may appear in a title of a book,
you can change it with another inexisting word.
Razvan

Can CASE match more than once ?

Hi
I'm tring to write a statement to analyse what orders were open on the first
day of each month from a system and return one data set with the months
listed and all the orders open during that month. eg.
Mon Order_No
Jan 001
Jan 002
Jan 003
Feb 002
Feb 003
Feb 004
The orders all have an open and closed date, so I want to check for each
month if the first of the month falls between the open and closed date of
each order.
I've set up a dummy database for testing - what I'd like to know is whether
CASE statements be made to match more than once :
SELECT MyNewField =
CASE
WHEN data1 = 1 THEN 'Is One'
WHEN data1 > 1 then 'Not One'
end,
data2
FROM APW_Test
my table is as follows
data1
1
2
3
4
So I would hope to see one result for the number 1 (Is One) and two results
for the remaining numbers because they match both case statements. However,
CASE seems to match the first statement and then stop for each record.
Is there a way I can achieve the result I want fairly simply ?
Thanks in advance.
AndrewHi ... I made a mistake in my logic. What I meant was

> WHEN data1 = 1 THEN 'Is One'
> WHEN data1 > 0 then 'Greater Than Zero'
so I expect 1 to appear twice. All else the same.
"Andrew Webb" <andrew.webb@.eme-med.co.uk> wrote in message
news:uMs5zBpuFHA.1256@.TK2MSFTNGP09.phx.gbl...
> Hi
> I'm tring to write a statement to analyse what orders were open on the
> first day of each month from a system and return one data set with the
> months listed and all the orders open during that month. eg.
> Mon Order_No
> Jan 001
> Jan 002
> Jan 003
> Feb 002
> Feb 003
> Feb 004
> The orders all have an open and closed date, so I want to check for each
> month if the first of the month falls between the open and closed date of
> each order.
> I've set up a dummy database for testing - what I'd like to know is
> whether CASE statements be made to match more than once :
> SELECT MyNewField =
> CASE
> WHEN data1 = 1 THEN 'Is One'
> WHEN data1 > 1 then 'Not One'
> end,
> data2
> FROM APW_Test
> my table is as follows
> data1
> 1
> 2
> 3
> 4
> So I would hope to see one result for the number 1 (Is One) and two
> results for the remaining numbers because they match both case statements.
> However, CASE seems to match the first statement and then stop for each
> record.
> Is there a way I can achieve the result I want fairly simply ?
> Thanks in advance.
> Andrew
>|||Andrew
you can simply use procedure/function to use if condition.
However post DDL,Sample data to help you better
Regards
R.D
"Andrew Webb" wrote:

> Hi ... I made a mistake in my logic. What I meant was
>
> so I expect 1 to appear twice. All else the same.
>
> "Andrew Webb" <andrew.webb@.eme-med.co.uk> wrote in message
> news:uMs5zBpuFHA.1256@.TK2MSFTNGP09.phx.gbl...
>
>|||On Fri, 16 Sep 2005 08:39:06 +0100, "Andrew Webb"
<andrew.webb@.eme-med.co.uk> wrote:

> Hi ... I made a mistake in my logic. What I meant was
>
> so I expect 1 to appear twice. All else the same.
CASE returns only the first match.|||Andrew,
It sounds to me like you need a JOIN operation, not a CASE expression.
Post the CREATE TABLE and INSERT statements for some specific
data if you want a more careful answer, but this might be close:
select data1, DisplayAnswer
from T join (
select 'equal 1' as TestCondition, 'Is One' as DisplayAnswer
union all
select 'above 1', 'Not One'
) C
on (
TestCondition = 'equal 1' and data1 = 1
) or (
TestCondition = 'above 1' and data1 > 1
)
If you select only from your 4-row table, with no other table
joined in, you cannot obtain a result that contains any row
more than once.
Steve Kass
Drew University
"Andrew Webb" <andrew.webb@.eme-med.co.uk> wrote in message
news:uMs5zBpuFHA.1256@.TK2MSFTNGP09.phx.gbl...
> Hi
> I'm tring to write a statement to analyse what orders were open on the
> first day of each month from a system and return one data set with the
> months listed and all the orders open during that month. eg.
> Mon Order_No
> Jan 001
> Jan 002
> Jan 003
> Feb 002
> Feb 003
> Feb 004
> The orders all have an open and closed date, so I want to check for each
> month if the first of the month falls between the open and closed date of
> each order.
> I've set up a dummy database for testing - what I'd like to know is
> whether CASE statements be made to match more than once :
> SELECT MyNewField =
> CASE
> WHEN data1 = 1 THEN 'Is One'
> WHEN data1 > 1 then 'Not One'
> end,
> data2
> FROM APW_Test
> my table is as follows
> data1
> 1
> 2
> 3
> 4
> So I would hope to see one result for the number 1 (Is One) and two
> results for the remaining numbers because they match both case statements.
> However, CASE seems to match the first statement and then stop for each
> record.
> Is there a way I can achieve the result I want fairly simply ?
> Thanks in advance.
> Andrew
>|||On Fri, 16 Sep 2005 08:30:00 +0100, Andrew Webb wrote:

>Hi
>I'm tring to write a statement to analyse what orders were open on the firs
t
>day of each month from a system and return one data set with the months
>listed and all the orders open during that month. eg.
>Mon Order_No
>Jan 001
>Jan 002
>Jan 003
>Feb 002
>Feb 003
>Feb 004
>The orders all have an open and closed date, so I want to check for each
>month if the first of the month falls between the open and closed date of
>each order.
Hi Andrew,
You could use a calendar table or a table of integers to get this. I'll
give an example with a table of integers that's made up "on the fly".
You can expand it as needed, or make a real table of integers (as
explained on http://www.aspfaq.com/show.asp?id=2516).
DECLARE @.StartDate smalldatetime
,@.EndDate smalldatetime
SET @.StartDate = '20050101' -- Should be first of the month
SET @.EndDate = '20051201'
SELECT DATEADD(month, Numbers.n, @.StartDate) AS Mon,
Orders.Order_No
FROM Orders
INNER JOIN (SELECT 1 UNION ALL SELECT 2 UNION ALL SELECT 3 UNION ALL
SELECT 4 UNION ALL SELECT 5 UNION ALL SELECT 6 UNION ALL
SELECT 9 UNION ALL SELECT 9 UNION ALL SELECT 10 UNION ALL
SELECT 11 UNION ALL SELECT 12 UNION ALL SELECT 13 UNION ALL
SELECT 14 UNION ALL SELECT 15) AS Numbers(n)
WHERE Orders.OpenDate < DATEADD(month, Numbers.n, @.StartDate)
AND Orders.CloseDate > DATEADD(month, Numbers.n, @.StartDate)
AND DATEADD(month, Numbers.n, @.StartDate) <= @.EndDate
ORDER BY Numbers.n, Orders.Order_No
Best, Hugo
--
(Remove _NO_ and _SPAM_ to get my e-mail address)