Sequential Required and Unordered Optional Elements

[echotrain]

I am having trouble building a schema that will validate sequential required elements followed by randomly ordered optional elements. This XML snippet is an example:

<Record>	 <RequiredElementOne>Must be first</RequiredElementOne>	 <RequiredElementTwo>Must be second</RequiredElementTwo>	 <OptionalElementOne>0-unbounded in any order</OptionalElementOne>	 <OptionalElementTwo>0-unbounded in any order</OptionalElementTwo></Record>

Either of the OptionalElements may or may not be present, in any quantity, in any order, and they must follow the RequiredElements. I have been trying variations on this:

<xs:element name="Record">    <xs:complexType>	  <xs:sequence>	    <xs:element ref="RequiredElementOne" />	    <xs:element ref="RequiredElementTwo" />	    <xs:group ref="OptionalGroup" />	  </xs:sequence>    </xs:complexType></xs:element><xs:element name="RequiredOne" /><xs:element name="RequiredTwo" /><xs:group name="OptionalGroup">    <xs:all>	  <xs:element name="OptionalElementOne" />	  <xs:element name="OptionalElementTwo" />    </xs:all></xs:group>

I keep bumping into the fact that xs:sequence and xs:all seem to be mutually exclusive. Any thoughts would be appreciated.

[boen_robot]

Yes, they are mutually exclusive, at least in XML Schema 1.0... you'd have to change your grammer so that you have just one optional element in a strict order, which may contain the rest in any order.Like:

<Record><RequiredElementOne>Must be first</RequiredElementOne><RequiredElementTwo>Must be second</RequiredElementTwo><OptionalElementHolder><OptionalElementOne>0-unbounded in any order</OptionalElementOne><OptionalElementTwo>0-unbounded in any order</OptionalElementTwo></OptionalElementHolder></Record>

and

<xs:element name="OptionalElementHolder" /><xs:group name="OptionalGroup"><xs:all><xs:element name="OptionalElementOne" /><xs:element name="OptionalElementTwo" /></xs:all></xs:group></xs:element>

[echotrain]

Thanks for your reply, boen_robot. I could live with that, but I couldn't get it to work. I could not get my parser to accept xs:all nested inside xs:group, however it did work nested in xs:complexType. So far, so good. But, now I find that the maxOccurs limit on elements within xs:all is 1. Since the minOccurs on my OptionalElements is 0 and the maxOccurs is unbounded, a Record element could have any number of ElementHolders that each hold only one element. As might be imagined, the production data structure is considerably more complex than these examples, and could easily become cluttered. I feel like this is on the right track. A little more brainstorming is needed.

[boen_robot]

My editor seems to say otherwise... the following schema is verified to be valid:

	<xs:element name="Record">		<xs:complexType>			<xs:sequence>				<xs:element name="RequiredElementOne"/>				<xs:element name="RequiredElementTwo"/>				<xs:element name="OptionalElementHolder" minOccurs="0">					<xs:complexType>						<xs:all>							<xs:element name="OptionalElementOne" minOccurs="0" maxOccurs="unbounded"/>							<xs:element name="OptionalElementTwo" minOccurs="0" maxOccurs="unbounded"/>						</xs:all>					</xs:complexType>				</xs:element>			</xs:sequence>		</xs:complexType>	</xs:element>

(optional elements can occur zero or more times in any order, as long as they're within OptionalElementHolder, which itself exists once as the last Record child element, or is not present at all)

[echotrain]

I am using Visual Studio, and consequently the MSXML engine: 0, 1 or left blank is all it will accept. I did some searching on this and found these references: http://connect.microsoft.com/VisualStudio/feedback/details/122890/xsd-xml-editor-incorrectly-complains-about-maxoccurs-1-in-xs-all-compositor That page refers to W3.org: http://www.w3.org/TR/xmlschema-1/#declare-contentModel Stylus Studio makes the same complaint, and I found this on their site: http://www.stylusstudio.com/xmldev/200110/post90460.html Thanks very much for your help. I feel a solution is lurking out there. More brainstorming!

[boen_robot]

I use Stylus Studio.... although I didn't tried to actually validate a document against it... just validated the schema itself, and there's no errors on that.

[echotrain]

OK, here is what I came up with. It's not pretty, but it validates. I'll try it out on the application Monday and report back.

 <xs:element name="Record">  <xs:complexType>   <xs:sequence>    <xs:element name="RequiredElementOne"/>    <xs:element name="RequiredElementTwo"/>    <xs:element name="Holder"  minOccurs="0">	 <xs:complexType>	  <xs:all>	   <xs:element name="OneHolder" minOccurs="0">	    <xs:complexType>		 <xs:sequence>		  <xs:element name="OptionalElementOne" minOccurs="0" maxOccurs="unbounded"/>		 </xs:sequence>	    </xs:complexType>	   </xs:element>	   <xs:element name="TwoHolder" minOccurs="0">	    <xs:complexType>		 <xs:sequence>		  <xs:element name="OptionalElementTwo" minOccurs="0" maxOccurs="unbounded"/>		 </xs:sequence>	    </xs:complexType>	   </xs:element>	  </xs:all>	 </xs:complexType>    </xs:element>   </xs:sequence>  </xs:complexType></xs:element>

[boen_robot]

OK, I tried to generate an XML file using my XML Schema, and it failed indeed. But I was reminded of an alternative approach... xs:choice can have quantifiers. Setting it to 0-unbounded along with its elements being fixed results in what I described above - a single optional element holder in which optional elements can appear zero or more times in any order:

	<xs:element name="Record">		<xs:complexType>			<xs:sequence>				<xs:element name="RequiredElementOne"/>				<xs:element name="RequiredElementTwo"/>				<xs:element name="OptionalElementHolder" minOccurs="0">					<xs:complexType>						<xs:choice minOccurs="0" maxOccurs="unbounded">							<xs:element name="OptionalElementOne"/>							<xs:element name="OptionalElementTwo"/>						</xs:choice>					</xs:complexType>				</xs:element>			</xs:sequence>		</xs:complexType>	</xs:element>

[echotrain]

xs:choice does not meet my requirement that both OptionalElementOne and OptionalElementTwo be permitted. It allows any number of one or the other, but not both.

[boen_robot]

No. The choice itself (not its elements) can occur between 0 and unlimited number of times. Each time you do make a choice, it's a choice between one OptionalElementOne or one OptionalElementTwo.The following XML validates with all validators using Stylus Studio:

<?xml version="1.0"?><Record xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xsi:noNamespaceSchemaLocation="TEST.xsd">    <RequiredElementOne>string by default</RequiredElementOne>    <RequiredElementTwo>string by default</RequiredElementTwo>    <!--Element OptionalElementHolder is optional-->    <OptionalElementHolder>        <OptionalElementTwo>string by default</OptionalElementTwo>        <OptionalElementTwo>string by default</OptionalElementTwo>        <OptionalElementOne>string by default</OptionalElementOne>        <OptionalElementOne>string by default</OptionalElementOne>        <OptionalElementTwo>string by default</OptionalElementTwo>        <OptionalElementTwo>string by default</OptionalElementTwo>        <OptionalElementOne>string by default</OptionalElementOne>        <OptionalElementOne>string by default</OptionalElementOne>        <OptionalElementTwo>string by default</OptionalElementTwo>        <OptionalElementOne>string by default</OptionalElementOne>        <OptionalElementOne>string by default</OptionalElementOne>        <OptionalElementOne>string by default</OptionalElementOne>        <OptionalElementOne>string by default</OptionalElementOne>    </OptionalElementHolder></Record>

[echotrain]

You are right, boen_robot. I will run my application against this model Monday and post back, but I see no reason it should not work.

[echotrain]

I have run my application with the xs:choice structure, and it does the job! It's clearer to me now, having seeing it in action - it's like restricting an element to a string or an integer, but in the case of xs:choice, the type is one of the element definitions. Thanks for the ideas, boen_robot, especially the one that worked.