Question

Problem:

Ajax suggest-search on [n] ingredients in recipes. That is: match recipes against multiple ingredients.

For instance: SELECT Recipes using 'flower', 'salt' would produce: 'Pizza', 'Bread', 'Saltwater' and so forth.

Tables:

Ingredients [     IngredientsID INT [PK],     IngredientsName VARCHAR ]  Recipes [     RecipesID INT [PK],     RecipesName VARCHAR ]  IngredientsRecipes [     IngredientsRecipesID INT [PK],     IngredientsID INT,     RecipesID INT ] 

Query:

SELECT     Recipes.RecipesID,     Recipes.RecipesName,     Ingredients.IngredientsID,     Ingredients.IngredientsName FROM     IngredientsRecipes      INNER JOIN Ingredients     ON IngredientsRecipes.IngredientsID = Ingredients.IngredientsID      INNER JOIN Recipes     ON IngredientsRecipes.RecipesID = Recipes.RecipesID WHERE     Ingredients.IngredientsName IN ('salt', 'water', 'flower') 

I am currently constructing my query using ASP.NET C# because of the dynamic nature of the WHERE clause.

I bites that I have to construct the query in my code-layer instead of using a stored procedure/pure SQL, which in theory should be much faster.

Have you guys got any thoughts on how I would move all of the logic from my code-layer to pure SQL, or at least how I can optimize the performance of what I’m doing?

I am thinking along the lines of temporary tables:

Step one: SELECT IngredientsID FROM Ingredients and INSERT INTO temp-table

Step two: SELECT RecipesName FROM Recipes joined with IngredientsRecipes joined with temp-table.IngredientsID

Answer 1

You have two options. If you’re using SQL Server 2008 (or Oracle) you can pass in a table value parameter.

If you’re using SQL Server 2005, you can use XML to simulate this capability

If you’re using something earlier than 2005, you need to concatenate the ids in a single string and create a UDF to parse them.