I have a table of data, and I allow people to add meta data to that table.
I give them an interface that allows them to treat it as though they're adding extra columns to the table their data is stored in, but I'm actually storing the data in another table.
Data Table
DataID
Data
Meta Table
DataID
MetaName
MetaData
So if they wanted a table that stored the data, the date, and a name, then I'd have the data in the data table, and the word "Date" in metaname, and the date in MetaData, and another row in the meta table with "Name" in metaname and the name in metadata.
I now need a query that takes the information from these tables and presents it as though coming from a single table with the two additional columns "Data" and "Name" so to the customer it would look like there's a single table with their custom columns:
MyTable
Data
Date
Name
Or, in other words, how do I go from here:
Data Table
DataID Data
1 Testing!
2 Hello, World!
Meta Table
DataID MetaName MetaData
1 Date 20081020
1 Name adavis
2 Date 20081019
2 Name mdavis
To here:
MyTable
Data Date Name
Testing! 20081020 adavis
Hello, World! 20081019 mdavis
Years ago when I did this in MySQL using PHP, I did two queries, the first to get the extra meta data, the second to join them all together. I'm hoping that modern databases have alternate methods of dealing with this.
Related to option 3 of this question.
-Adam
You want to pivot each of your name-value pair rows in the MyTable... Try this sql:
DECLARE @Data TABLE (
DataID INT IDENTITY(1,1) PRIMARY KEY,
Data VARCHAR(MAX)
)
DECLARE @Meta TABLE (
DataID INT ,
MetaName VARCHAR(MAX),
MetaData VARCHAR(MAX)
)
INSERT INTO @Data
SELECT 'Data'
INSERT INTO @Meta
SELECT 1, 'Date', CAST(GetDate() as VARCHAR(20))
UNION
SELECT 1, 'Name', 'Joe Test'
SELECT * FROM @Data
SELECT * FROM @Meta
SELECT
D.DataID,
D.Data,
MAX(CASE MetaName WHEN 'Date' THEN MetaData ELSE NULL END) as Date,
MAX(CASE MetaName WHEN 'Name' THEN MetaData ELSE NULL END) as Name
FROM
@Meta M
JOIN @Data D ON M.DataID = D.DataID
GROUP BY
D.DataID,
D.Data
SELECT DataTable.Data AS Data, MetaTable.MetaData AS Date, MetaTable.MetaName AS Name
FROM DataTable, MetaTable
WHERE DataTable.DataID = MetaTable.DataID
You'll probably want to add an additional clause (AND Data = 'some value') to return the rows the user is interested in.
AFAIK, you can do this on the server-side only with a dynamic SQL
stored procedure.
Effectively the code you want to generate dynamically is:
SELECT [Data Table].*
,[MyTable Date].MetaData
,[MyTable Name].MetaData
FROM [Data Table]
LEFT JOIN [MyTable] AS [MyTable Date]
ON [MyTable Date].DataID = [Data Table].DataID
AND [MyTable Date].MetaName = 'Date'
LEFT JOIN [MyTable] AS [MyTable Name]
ON [MyTable Name].DataID = [Data Table].DataID
AND [MyTable Name].MetaName = 'Name'
And here's code to do it:
DECLARE @sql AS varchar(max)
DECLARE @select_list AS varchar(max)
DECLARE @join_list AS varchar(max)
DECLARE @CRLF AS varchar(2)
DECLARE @Tab AS varchar(1)
SET @CRLF = CHAR(13) + CHAR(10)
SET @Tab = CHAR(9)
SELECT @select_list = COALESCE(@select_list, '')
+ @Tab + ',[MyTable_' + PIVOT_CODE + '].[MetaData]' + @CRLF
,@join_list = COALESCE(@join_list, '')
+ 'LEFT JOIN [MyTable] AS [MyTable_' + PIVOT_CODE + ']' + @CRLF
+ @Tab + 'ON [MyTable_' + PIVOT_CODE + '].DataID = [DataTable].DataID' + @CRLF
+ @Tab + 'AND [MyTable_' + PIVOT_CODE + '].MetaName = ''' + PIVOT_CODE + '''' + @CRLF
FROM (
SELECT DISTINCT MetaName AS PIVOT_CODE
FROM [MyTable]
) AS PIVOT_CODES
SET @sql = 'SELECT [DataTable].*' + @CRLF
+ @select_list
+ 'FROM [DataTable]' + @CRLF
+ @join_list
PRINT @sql
--EXEC (@sql)
You could use a similar dynamic technique using the CASE
statement example to perform the pivot.