SQL - XML - 在子查询中创建多个同名节点

Question

I'm having an issue, when generating an XML file from SQL. Generally it works fine, but now the issue is, that I want to display mutliple nodes with the same name <cbc:Note> in a subquery. I know I can force a split with a null column in between, but then it generates a new InvoiceLine

The original query has more than 1000 lines of code, so I will just post the data, which is needed.

The query starts this way:

WITH XMLNAMESPACES (
                        'urn:oasis:names:specification:ubl:schema:xsd:CommonExtensionComponents-2' as ext,
                        'urn:oasis:names:specification:ubl:schema:xsd:CommonBasicComponents-2' as cbc,
                        'urn:oasis:names:specification:ubl:schema:xsd:CommonAggregateComponents-2' as cac,
                        'http://uri.etsi.org/01903/v1.3.2#' as xades,
                        'http://www.w3.org/2001/XMLSchema-instance' as xsi,
                        'http://www.w3.org/2000/09/xmldsig#' as ds
                    )   

SELECT
    @XMLData = xmldat.xmldataCol 
FROM
(
    SELECT
        (

        SELECT
            -- HIER XML Daten generieren
            ''                                                              as 'ext:UBLExtensions',
            ''                                                              as 'ext:UBLExtensions/ext:UBLExtension',
            ''                                                              as 'ext:UBLExtensions/ext:UBLExtension/ext:ExtensionContent',
            '2.1'                                                           as 'cbc:UBLVersionID',
            'TR1.2'                                                         as 'cbc:CustomizationID',
            -- evtl. Kommentar für geschart = 1
            case 
            when v2.geschart != 1 then 'IHRACAT'
            else '###PROFILE###' end                                        as 'cbc:ProfileID',
            '###TRANSACTION_NO###'                                          as 'cbc:ID',
            'false'                                                         as 'cbc:CopyIndicator',
            ....

At some point in the code it generates multiple invoice lines. In an invoice line, I need to have multiple <cbc:Note> nodes. This is the code:

        ...
        case when v2.RECINFW = 1 then
            cast(v2.kurs as decimal(18,4))              
        else
            null
        end                                         as 'cac:PricingExchangeRate/cbc:CalculationRate',
        /* 
            Invoice Lines
            | | | | | | | | | | | | | | | | |
            v v v v v v v v v v v v v v v v v
        */
        (
            SELECT 
                vp2.pos_nr                                                      as 'cac:InvoiceLine/cbc:ID',
                'PalWeight:' + cast(cast(isnull(vp2.gew_pal,'') as decimal(18,4)) as nvarchar(50))  as 'cbc:Note',
                'PackUnit:' + cast(cast(isnull(a.VERP_EIN,'') as decimal(18,4)) as nvarchar(50))            as 'cbc:Note',
                case when vp2.DRUCKFLG=4 then cast(isnull(vp2.BEMERK,'') as nvarchar(max)) else null end    as 'cbc:Note',
                ''                                                                                          as 'cbc:SomeOtherField',
                ...
            FROM 
                vorgpos2 vp2 (nolock)
            inner join
                artikel a (nolock) on vp2.ARTKENN = a._KENN
            WHERE 
                vp2.vorgang2 = v2._kenn FOR XML path (''), root('REPLACEINVOICELINEREPLACE'), type
        ),

When I put a null in between the cbc:Notes columns,

'PalWeight:' + cast(cast(isnull(vp2.gew_pal,'') as decimal(18,4)) as nvarchar(50))  as 'cbc:Note',
null,
'PackUnit:' + cast(cast(isnull(a.VERP_EIN,'') as decimal(18,4)) as nvarchar(50))            as 'cbc:Note',
null,
case when vp2.DRUCKFLG=4 then cast(isnull(vp2.BEMERK,'') as nvarchar(max)) else null end    as 'cbc:Note',

it breaks the complete invoice line (stop one and create a new one):

<cac:InvoiceLine>
    <cbc:ID>1</cbc:ID>
    <cbc:Note>PalWeight:1000.0000</cbc:Note>
</cac:InvoiceLine>
<cac:InvoiceLine>
    <cbc:Note>PackUnit:25.0000</cbc:Note>
</cac:InvoiceLine>
<cac:InvoiceLine>
    <cbc:InvoicedQuantity unitCode="KGM">100.0000</cbc:InvoicedQuantity>
    ...

Wanted result:

<cac:InvoiceLine>
    <cbc:ID>1</cbc:ID>
    <cbc:Note>PalWeight:1000.0000</cbc:Note>
    <cbc:Note>PackUnit:25.0000</cbc:Note>
    <cbc:InvoicedQuantity unitCode="KGM">100.0000</cbc:InvoicedQuantity>
    ...

If you wonder about the missing REPLACEINVOICELINEREPLACE root element in the subquery. This is being replaced afterwards. Has anyone a hint or a solution for this problem? Many thanks in advance!

Answer 1

SQL XML is all kinds of weird. You might want to try something like the following (note the null as 'x' columns):

WITH XMLNAMESPACES (
  'urn:oasis:names:specification:ubl:schema:xsd:CommonBasicComponents-2' as cbc,
  'urn:oasis:names:specification:ubl:schema:xsd:CommonAggregateComponents-2' as cac
)
SELECT
  pos_nr                                                      as 'cbc:ID',
  'PalWeight:' + cast(cast(isnull(gew_pal,'') as decimal(18,4)) as nvarchar(50))  as 'cbc:Note',
  null as 'x',
  'PackUnit:' + cast(cast(isnull(VERP_EIN,'') as decimal(18,4)) as nvarchar(50))            as 'cbc:Note',
  null as 'x',
  case when DRUCKFLG=4 then cast(isnull(BEMERK,'') as nvarchar(max)) else null end    as 'cbc:Note'
FROM (values
  (1, 1000.0, 25.0, 4, 100.0)
) dat (pos_nr, gew_pal, verp_ein, druckflg, bemerk)
FOR XML path ('cac:InvoiceLine'), root('foo'), type

Which yields:

<foo xmlns:cac="urn:oasis:names:specification:ubl:schema:xsd:CommonAggregateComponents-2" xmlns:cbc="urn:oasis:names:specification:ubl:schema:xsd:CommonBasicComponents-2">
  <cac:InvoiceLine>
    <cbc:ID>1</cbc:ID>
    <cbc:Note>PalWeight:1000.0000</cbc:Note>
    <cbc:Note>PackUnit:25.0000</cbc:Note>
    <cbc:Note>100.0</cbc:Note>
  </cac:InvoiceLine>
</foo>

Credit where credit is due: for xml path() concatenates texts, how to get 2 elements with the same name?

Answer 2

Note: pay attention to casting an empty string '' to decimal, you'll get an error

select 'PalWeight:' + cast(cast(isnull(null,'') as decimal(18,4)) as nvarchar(50)) 
go
--concat is more flexible with null
select concat('PalWeight:', null) 
go

The xml part:

--edited this part, to include the attribute on a column (the attribute to a following element was apparently the culprit and i got confused as to how multi InvoiceLines could happen). 
--Just adjust the other xml generation options accordingly.

select
pos_nr as 'InvoiceLine/ID',
null as 'InvoiceLine',
gew_pal as 'InvoiceLine/Note',
null as 'InvoiceLine',
verp_ein as 'InvoiceLine/Note',
null as 'InvoiceLine',
DRUCKFLG as 'InvoiceLine/Note',
someothercol3 as 'InvoiceLine/someothercol1/@attributeofthefollowingcolumncol1', --<-- reason of confusion
someothercol1 as 'InvoiceLine/someothercol1',
someothercol2 as 'InvoiceLine/someothercol2' 
from 
(
values(1, 'a1', 'b1', 'c1', 'd1', 'e1', 'f1'),(2, 'a2', 'b2', null, 'd2', null, 'f2'),(3, null, 'b3', 'c3', 'd3', null, null)
) as t(pos_nr, gew_pal, verp_ein, DRUCKFLG, someothercol1, someothercol2, someothercol3)
for xml path('');


select
pos_nr as 'InvoiceLine/ID',
(select 
    gew_pal as 'Note',
    null,
    verp_ein as 'Note',
    null,
    DRUCKFLG as 'Note',
    someothercol1 as 'someothercol1',
    someothercol2 as 'someothercol2' 
for xml path(''), type
) as 'InvoiceLine/*'
from 
(
values(1, 'a1', 'b1', 'c1', 'd1', 'e1', 'f1'),(2, 'a2', 'b2', null, 'd2', null, 'f2'),(3, null, 'b3', 'c3', 'd3', null, null)
) as t(pos_nr, gew_pal, verp_ein, DRUCKFLG, someothercol1, someothercol2, someothercol3)
for xml path('');


select
pos_nr as 'ID',
(select 
    gew_pal as 'Note',
    null,
    verp_ein as 'Note',
    null,
    DRUCKFLG as 'Note',
    someothercol1 as 'someothercol1',
    someothercol2 as 'someothercol2' 
for xml path(''), type
) 
from 
(
values(1, 'a1', 'b1', 'c1', 'd1', 'e1', 'f1'),(2, 'a2', 'b2', null, 'd2', null, 'f2'),(3, null, 'b3', 'c3', 'd3', null, null)
) as t(pos_nr, gew_pal, verp_ein, DRUCKFLG, someothercol1, someothercol2, someothercol3)
for xml path('InvoiceLine');

Answer 3

You can either use another sub-select (if your data is set based, or an element closer in between:

        SELECT 
            vp2.pos_nr                                                      as 'cac:InvoiceLine/cbc:ID',
            'PalWeight:' + cast(cast(isnull(vp2.gew_pal,'') as decimal(18,4)) as nvarchar(50))  as 'cac:InvoiceLine/cbc:Note',
            NULL as'cac:InvoiceLine',
            'PackUnit:' + cast(cast(isnull(a.VERP_EIN,'') as decimal(18,4)) as nvarchar(50))            as 'cac:InvoiceLine/cbc:Note',
            NULL as'cac:InvoiceLine',
            case when vp2.DRUCKFLG=4 then cast(isnull(vp2.BEMERK,'') as nvarchar(max)) else null end    as 'cac:InvoiceLine/cbc:Note',
            ''                                                                                          as 'cac:InvoiceLine/cbc:SomeOtherField',
            ...
        FROM 

The idea in short:

The engine opens the <cac:InvoiceLine> and then opens the <cbc:ID> element to insert the vb2.pos_nr as text() -node.
The next line tells the engine the path 'cac:InvoiceLine/cbc:Note' . The engine thinks: "Oh, nothing more for <cbc:ID> , let's close this element. Ah, we continue with <cac:InvoiceLine> , which is still open. Let's insert a <cbc:Note> " .
Now the magic happes: The eninge thinks: "Ok, we have to close <cbc:Node> , but we continue within <cac:InvoiceLine> . Oops, nothing to insert... Never mind, let's pick the next. Okay, we are still within <cac:InvoiceLine> but we have to open a <cbc:Node> element."
And so on...