Question

I am creating a small search function on my site, this enables the search for articles in the system. Each article has a set of keywords associated to it and these keywords are stored inside a SQL Server database.

This is the table:

CREATE TABLE [dbo].[SearchWords] (
    [ID] [int] IDENTITY(1,1) NOT NULL,
    [ArticleID] [int] NOT NULL,
    [SearchWord] [nvarchar](20) NOT NULL,
    CONSTRAINT [PK_SearchWords] PRIMARY KEY CLUSTERED 
        ([ID] ASC) 
        WITH (PAD_INDEX  = OFF, STATISTICS_NORECOMPUTE = OFF, 
              IGNORE_DUP_KEY = OFF, ALLOW_ROW_LOCKS = ON, 
              ALLOW_PAGE_LOCKS  = ON)
        ON [PRIMARY]
) ON [PRIMARY]

Each article can have an unlimited amount of keywords. Now my problem is the search itself.
When for example the user types:

France actors

I want the system to find all articles (just once DISTINCT) with the keywords France and actors. I am passing the search criteria as a varchar (separated with space) to the stored procedure. Then i am splitting the words up with the following function: (Erland Sommarskog) http://www.sommarskog.se/arrays-in-sql-2005.html#iter-list-of-strings

How do i then match the criteria words against the search words and only get the distinct article ids?

I am working with something like this, just that i cant understand how to match all the keywords. This method works if just one keyword is entered. If the user enters multiple, then it doesn’t return anything even if the article has all involved keywords.

declare @temp nvarchar(50)
set @temp = 'France actors'

SELECT DISTINCT Article.ArticleID 
FROM Article
INNER JOIN SearchWords 
    ON Article.ArticleID = SearchWords.ArticleID
JOIN iter_charlist_to_tbl(@temp, DEFAULT) s 
    ON SearchWords.SearchWord = s.nstr

Any ideas?

Answer 1

First off, your linked UDF’s default delimiter is the , character, not space. So, using the default delimiter as you are you’re getting a single row back with both words in it. (Debugging hint: when something isn’t working right, take it apart. In this case, you should have done a select * from UDF(@temp, DEFAULT) to see if the table looked correct.)

Assuming you want to keep using that UDF and you want articles that match any of the search terms (but not necessarily all), something along these lines should be correct:

declare @temp nvarchar(50)
set @temp = 'France actors'

SELECT DISTINCT 
  a.ArticleID 
FROM 
  Article a
  JOIN SearchWords sw ON a.ArticleID = sw.ArticleID
WHERE
  exists (
    select 
      1 
    from 
      iter_charlist_to_tbl(@temp, ' ') s 
    where
      s.nstr = sw.SearchWord
  )

Your inner join method probably should work as well if you change the parameters to the UDF.