TSQL query xml output

Question

I have a query that I am running for some search results on my page. The stored procedure looks like this:

BEGIN
SET NOCOUNT ON;
BEGIN
    SELECT *
    FROM   (SELECT id,
                   data,
                   dataType,
                   dataLocation,
                   tag,
                   whoAdded,
                   whenAdded,
                   notes,
                   ROW_NUMBER() OVER (PARTITION BY data ORDER BY whenAdded DESC) AS rn
            FROM   Tags_Accounts
            WHERE  tag IN (SELECT *
                           FROM   dbo.splitstring (@tags))) AS a
    WHERE  rn = 1
    FOR    XML PATH ('results'), TYPE, ELEMENTS, ROOT ('root');
END
END

This will go through my records that looks like the following:

Now, If I were to search for the tag stuff, paypal it should return 2 results as the partition is on the data column.

The XML output is as follows for the results:

            <root>
              <results>
                <id>43</id>
                <data>123333</data>
                <dataType>1</dataType>
                <dataLocation>AF</dataLocation>
                <tag>paypal</tag>
                <whoAdded>chussey</whoAdded>
                <whenAdded>2013-11-22T11:01:50.117</whenAdded>
                <rn>1</rn>
              </results>
              <results>
                <id>41</id>
                <data>12345</data>
                <dataType>1</dataType>
                <dataLocation>AF</dataLocation>
                <tag>paypal</tag>
                <whoAdded>chussey</whoAdded>
                <whenAdded>2013-11-22T10:59:39.277</whenAdded>
                <rn>1</rn>
              </results>
              <results>
                <id>50</id>
                <data>RGG</data>
                <dataType>2</dataType>
                <dataLocation>AF</dataLocation>
                <tag>stuff</tag>
                <whoAdded>chussey</whoAdded>
                <whenAdded>2013-11-22T22:25:41.393</whenAdded>
                <rn>1</rn>
              </results>
            </root>

It included the 2 results for PayPal as the data was different in the data col. It returned 1 record for stuff as the data was already found matching in the results for the tag PayPal.

My question: is there any way to do exactly what it's doing now but in the <Tag></Tag> include all of the tags that match the data in the data column?

For example, because the tag stuff was found with the other tag Paypal it would include them together.

The goal out of this is to find any data that contains the tag searched. If its found and there are more than 1 tag for that piece of data, it will include all the tags with it.

<results>
    <id>50</id>
    <data>RGG</data>
    <dataType>2</dataType>
    <dataLocation>AF</dataLocation>
    <tag>stuff, testing</tag>
    <whoAdded>chussey</whoAdded>
    <whenAdded>2013-11-22T22:25:41.393</whenAdded>
    <rn>1</rn>
</results>

Answer 1

I think this is doing what you want:

WITH OrderedTags
AS
(
    SELECT  id,
            data,   
            dataType,
            dataLocation,
            CAST(tag as NVARCHAR(max)) as tag,
            whoAdded,
            whenAdded,
            notes,
            ROW_NUMBER() OVER (PARTITION BY data ORDER BY tag, id) AS rn
    FROM    Tags_Accounts T
)
,RecurseTags (data, tags, rn)
AS
(
--Anchor
    SELECT  data,
            tag as tags,
            rn
    FROM    OrderedTags T
    WHERE   rn = 1
    UNION ALL
--Recurse
    SELECT  R.data,
            tags + N', ' + tag,
            T.rn
    FROM    RecurseTags R
    JOIN    OrderedTags T
        ON  T.data = R.data
        AND T.rn = R.rn + 1
)
,CommaSeparatedTags(data, tags)
AS
(
    SELECT  data, tags
    FROM
    (
        SELECT  data,
                tags,
                ROW_NUMBER() OVER (PARTITION BY data ORDER BY rn DESC) AS rn
        FROM    RecurseTags
    ) T
    WHERE rn = 1
)
,RequiredRows
AS
(
    SELECT * FROM(
    SELECT  A.id,
            A.data, 
            A.dataType,
            A.dataLocation,
            C.tags as tag,
            A.whoAdded,
            A.whenAdded,
            A.notes,
            ROW_NUMBER() OVER (PARTITION BY A.data ORDER BY A.whenAdded DESC) AS rn
    FROM   OrderedTags A
    JOIN   dbo.splitstring (@tags) T
        ON T.Value = A.tag
    JOIN    CommaSeparatedTags C
        ON C.data = A.data) A
    WHERE rn = 1
)
SELECT * FROM RequiredRows
FOR    XML PATH ('results'), TYPE, ELEMENTS, ROOT ('root')

Basically, it orders the tags by data, generates a consolidated table, and then joins your original query to that...