交叉应用返回太多行

Question

DDL:

CREATE TABLE [testXML]
(
    [scheduleid] [uniqueidentifier] primary key,
    [XMLData1] [xml] NULL
)

INSERT INTO testXML ([scheduleid],XMLData1) 
VALUES ('88888888-DDDD-4444-AAAA-666666666666','<ArrayOfRDData xmlns:xsd="http://www.w3.org/2001/XMLSchema" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance">
  <RDData  h="Title"  o="2017-11-02T16:00:00Z" p="212" q="000000cb-0000-0000-0000-000000000000" an="203"  >
    <rps>
      <rp s="00a566e2-0000-0000-0000-000000000000"  ag="1" i="0" j="0" ah="2018-01-10T17:00:00Z" >
        <piData programId="00a566e2-0000-0000-0000-000000000000" al="0" />
        <res>
          <re o="2018-01-10T17:00:00Z"  p="212" q="000000cb-0000-0000-0000-000000000000" >
            <riData av="false"  az="201" />
          </re>
        </res>
      </rp>
      <rp s="00a5860a-0000-0000-0000-000000000000" ag="1" i="0" j="0" ah="2018-01-26T17:00:00Z" >
        <piData programId="00a5860a-0000-0000-0000-000000000000" al="1" />
        <res>
          <re o="2018-01-26T17:00:00Z"  p="212" q="000000cb-0000-0000-0000-000000000000" >
            <riData av="false"  az="201" />
          </re>
        </res>
      </rp>
      <rp s="00a595c4-0000-0000-0000-000000000000" ag="0" i="0" j="0" ah="2018-01-31T17:00:00Z" >
        <piData programId="00a595c4-0000-0000-0000-000000000000" al="2"  />
        <res>
          <re o="2018-01-31T17:00:00Z"  p="212" q="000000cb-0000-0000-0000-000000000000"  />
        </res>
      </rp>
      <rp s="00a595c0-0000-0000-0000-000000000000" ag="1" i="0" j="0" ah="2018-01-29T17:00:00Z" >
        <piData programId="00a595c0-0000-0000-0000-000000000000" al="3"  />
        <res>
          <re o="2018-01-29T17:00:00Z" p="212" q="000000cb-0000-0000-0000-000000000000"  >
            <riData av="false"  az="180"  />
          </re>
        </res>
      </rp>
    </rps>
  </RDData>
</ArrayOfRDData>')

Query:

SELECT 
    [scheduleid],
    StationID_q = ARD3.res.value('@q', 'varchar(max)'),
    ProgramID_s = ARD2.ag.value('@s', 'varchar(max)'),
    StartTime_o = ARD3.res.value('@o', 'datetime') 
FROM
    [DVR_0601].[dbo].testXML Sch 
CROSS APPLY
    Sch.XMLData1.nodes('/ArrayOfRDData/RDData/rps') AS AoD(RDData) 
CROSS APPLY
    AoD.RDData.nodes('rp') AS ARD2(ag) 
CROSS APPLY
    AoD.RDData.nodes('rp/res/re') AS ARD3(res) 
WHERE
    ISNULL( ARD2.ag.value('@ag', 'int'), 0) = 1               

Output:

StationID_q                          ProgramID_s                       StartTime_o

000000cb-0000-0000-0000-000000000000 00a566e2-0000-0000-0000-000000000000   2018-01-10 17:00:00.000
000000cb-0000-0000-0000-000000000000    00a566e2-0000-0000-0000-000000000000    2018-01-26 17:00:00.000
000000cb-0000-0000-0000-000000000000    00a566e2-0000-0000-0000-000000000000    2018-01-31 17:00:00.000
000000cb-0000-0000-0000-000000000000    00a566e2-0000-0000-0000-000000000000    2018-01-29 17:00:00.000
000000cb-0000-0000-0000-000000000000    00a5860a-0000-0000-0000-000000000000    2018-01-10 17:00:00.000
000000cb-0000-0000-0000-000000000000    00a5860a-0000-0000-0000-000000000000    2018-01-26 17:00:00.000
000000cb-0000-0000-0000-000000000000    00a5860a-0000-0000-0000-000000000000    2018-01-31 17:00:00.000
000000cb-0000-0000-0000-000000000000    00a5860a-0000-0000-0000-000000000000    2018-01-29 17:00:00.000
000000cb-0000-0000-0000-000000000000    00a595c0-0000-0000-0000-000000000000    2018-01-10 17:00:00.000
000000cb-0000-0000-0000-000000000000    00a595c0-0000-0000-0000-000000000000    2018-01-26 17:00:00.000
000000cb-0000-0000-0000-000000000000    00a595c0-0000-0000-0000-000000000000    2018-01-31 17:00:00.000
000000cb-0000-0000-0000-000000000000    00a595c0-0000-0000-0000-000000000000    2018-01-29 17:00:00.000

Required output:

StationID_q ProgramID_s StartTime_o
000000cb-0000-0000-0000-000000000000    00a566e2-0000-0000-0000-000000000000    2018-01-10 17:00:00.000
000000cb-0000-0000-0000-000000000000    00a5860a-0000-0000-0000-000000000000    2018-01-26 17:00:00.000
000000cb-0000-0000-0000-000000000000    00a595c0-0000-0000-0000-000000000000    2018-01-29 17:00:00.000

I am getting cross joins between the rows of <rp> data.

Also note if ag="0" I want to skip that data, it does, just it still joins that row. I'm not sure how to join the <rp> to <re> or if it's at all possible.

Answer 1

You're using too many times CROSS APPLY.

SELECT --[scheduleid],
       StationID_q = ARD3.res.value('@q', 'varchar(max)'),
       ProgramID_s = ARD2.ag.value('@s', 'varchar(max)'),
       StartTime_o = ARD3.res.value('@o', 'datetime')
FROM   testXML Sch
---------------- all rp nodes 
CROSS APPLY Sch.XMLData1.nodes('/ArrayOfRDData/RDData/rps/rp') AS ARD2(ag)
---------------- all inner res/re nodes from rp tag 
CROSS APPLY ag.nodes('res/re') AS ARD3(res) 
WHERE       ISNULL(ARD2.ag.value('@ag', 'int'), 0) = 1

\nStationID_q |  ProgramID_s |  StartTime_o         \n:----------------------------------- |  :----------------------------------- |  :------------------ \n000000cb-0000-0000-0000-000000000000 |  00a566e2-0000-0000-0000-000000000000 |  10/01/2018 17:00:00 \n000000cb-0000-0000-0000-000000000000 |  00a5860a-0000-0000-0000-000000000000 |  26/01/2018 17:00:00 \n000000cb-0000-0000-0000-000000000000 |  00a595c0-0000-0000-0000-000000000000 |  29/01/2018 17:00:00 \n

dbfiddle here

Answer 2

This approach reads the meta data out of the XML directly and needs only one single CROSS APPLY and no WHERE (the filter is included into nodes() in the predicate ):

SELECT testXML.XMLData1.value('(/ArrayOfRDData/RDData/@q)[1]','uniqueidentifier') AS StationID
      ,a.rps.value('@s','uniqueidentifier') AS ProgramID
      ,a.rps.value('(res/re/@o)[1]','datetime') AS StartTime
FROM testXML
CROSS APPLY testXML.XMLData1.nodes('/ArrayOfRDData/RDData/rps/rp[@ag!=0]') AS a(rps);