sql server, defined xml schema to column

Question

im trying to add the xml schema to the columns of my table. what i would to do is:

• extract schema from a query
• add the schema to column
• insert xml to this column

my problem is how to exclude some options in the schema, like the possibility to don t check empty elements.

so, for example:

create table #t1 (c1 int not null,c2 int null)

  select CONVERT(xml,(select *
    from #t1
    where 0=1
    for xml raw,elements,XMLSCHEMA('XXX')
    ))

this get me

<xsd:schema xmlns:xsd="http://www.w3.org/2001/XMLSchema" xmlns:sqltypes="http://schemas.microsoft.com/sqlserver/2004/sqltypes" targetNamespace="XXX" elementFormDefault="qualified">
  <xsd:import namespace="http://schemas.microsoft.com/sqlserver/2004/sqltypes" schemaLocation="http://schemas.microsoft.com/sqlserver/2004/sqltypes/sqltypes.xsd" />
  <xsd:element name="row">
    <xsd:complexType>
      <xsd:sequence>
        <xsd:element name="c1" type="sqltypes:int" />
        <xsd:element name="c2" type="sqltypes:int" minOccurs="0" />
      </xsd:sequence>
    </xsd:complexType>
  </xsd:element>
</xsd:schema>

how can define all elements with "minOccurs="0""??

thank you!

Answer 1

There is - for sure! - no easy going ! The engine is obliged to return the schema as it was found. You might take the schema as an XML and modify it, you might use a string based approach (or some kind of regex ). The following is a quirky, but working dynamic approach:

The procedure sp_describe_first_result_set is used to get meta data of an unknown result set. You can store its return in a table.

Using the information you can create a copy of your table, where all columns are nullable. Use this new table to get your schema:

create table #FirstResult(is_hidden BIT
                         ,column_ordinal INT
                         ,name NVARCHAR(MAX)
                         ,is_nullable BIT
                         ,system_type_id BIT
                         ,system_type_name VARCHAR(MAX)
                         ,max_length INT
                         ,[precision] INT
                         ,scale INT
                         ,collation_name VARCHAR(MAX)
                         ,user_type_id INT
                         ,user_type_database VARCHAR(MAX)
                         ,user_type_schema VARCHAR(MAX)
                         ,user_type_name VARCHAR(MAX)
                         ,assembly_qualified_type_name VARCHAR(MAX)
                         ,xml_collection_id VARCHAR(MAX)
                         ,xml_collection_database VARCHAR(MAX)
                         ,xml_collection_schema VARCHAR(MAX)
                         ,xml_collection_name VARCHAR(MAX)
                         ,is_xml_document BIT
                         ,is_case_sensitive BIT
                         ,is_fixed_length_clr_type BIT
                         ,source_server VARCHAR(MAX)
                         ,source_database VARCHAR(MAX)
                         ,source_schema VARCHAR(MAX)
                         ,source_table VARCHAR(MAX)
                         ,source_column VARCHAR(MAX)
                         ,is_identity_column BIT
                         ,is_part_of_unique_key BIT
                         ,is_updateable BIT
                         ,is_computed_column BIT
                         ,is_sparse_column_set BIT
                         ,ordinal_in_order_by_list INT
                         ,order_by_is_descending BIT
                         ,order_by_list_length INT
                         ,tds_type_id INT
                         ,tds_length INT
                         ,tds_collation_id INT
                         ,tds_collation_sort_id INT);

--Your use case

create table #t1 (c1 int not null,c2 int null);
GO

INSERT INTO #FirstResult
exec sp_describe_first_result_set @tsql=N'SELECT * FROM #t1';

DECLARE @cmd NVARCHAR(MAX)=
(
SELECT N'DECLARE @tmp TABLE('
+
STUFF(
(
SELECT N',' + name + ' ' + system_type_name + N' NULL' 
FROM #FirstResult
FOR XML PATH(''),TYPE
).value('text()[1]','nvarchar(max)'),1,1,'')
+N'); 
SELECT * FROM @tmp AS SomeTable FOR XML RAW,ELEMENTS,XMLSCHEMA(''XXX'')'
);
EXEC (@cmd);

GO
drop table #t1;
drop table #FirstResult;
GO

The result

<xsd:schema targetNamespace="XXX" xmlns:xsd="http://www.w3.org/2001/XMLSchema" xmlns:sqltypes="http://schemas.microsoft.com/sqlserver/2004/sqltypes" elementFormDefault="qualified">
  <xsd:import namespace="http://schemas.microsoft.com/sqlserver/2004/sqltypes" schemaLocation="http://schemas.microsoft.com/sqlserver/2004/sqltypes/sqltypes.xsd" />
  <xsd:element name="row">
    <xsd:complexType>
      <xsd:sequence>
        <xsd:element name="c1" type="sqltypes:int" minOccurs="0" />
        <xsd:element name="c2" type="sqltypes:int" minOccurs="0" />
      </xsd:sequence>
    </xsd:complexType>
  </xsd:element>
</xsd:schema>  

Answer 2

If you don't mind the difference between minOccurs="0" and nillable ( http://www.dimuthu.org/blog/2008/08/18/xml-schema-nillabletrue-vs-minoccurs0/comment-page-1/ ) then you can do

create table #t1 (c1 int not null,c2 int null)

  select CONVERT(xml,(select *
    from #t1
    where 0=1
    for xml raw,elements xsinil,XMLSCHEMA('XXX')
    ))