Wednesday, November 21, 2012

Key, KeyRef and Unique in XSD


xs:key and xs:KeyRef are two XSD elements that defines the association of two elements in a XML document.  It is similar to the primary key and foreign key in the database.  It forces the association constraints between the elements.

xs:unique is XSD element that forces the uniqueness of value of specified XML element or attribute.

Here I will demonstrate the use of these XSD element in the below example.

<?xml version="1.0" encoding="UTF-8"?>
<xs:schema xmlns:xs="http://www.w3.org/2001/XMLSchema" xmlns:tns="http://www.toic.com/cdm" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" targetNamespace="http://www.toic.com/cdm" elementFormDefault="qualified" attributeFormDefault="unqualified">
            <xs:element name="Passengers">
                        <xs:complexType>
                                    <xs:choice maxOccurs="unbounded">
                                                <xs:element name="Passenger" type="tns:PassengerType"/>
                                                <xs:element name="Infant" type="tns:InfantType"/>
                                    </xs:choice>
                        </xs:complexType>
                        <xs:unique name=" PassengerInfantUniqueSequnceNo ">
                                    <xs:selector xpath="tns:Passenger | tns:Infant"/>
                                    <xs:field xpath="@SequenceNo"/>
                        </xs:unique>
                                <xs:unique name="PassengerInfantUniqueID">
                                                <xs:selector xpath="tns:Passenger | tns:Infant"/>
                                                <xs:field xpath="@ID"/>
                                </xs:unique>    
                        <xs:key name="PassengerIdKey">
                                    <xs:selector xpath="tns:Passenger"/>
                                    <xs:field xpath="@ID"/>
                        </xs:key>
                        <xs:key name="InfantIdKey">
                                    <xs:selector xpath="tns:Infant"/>
                                    <xs:field xpath="@ID"/>
                        </xs:key>
                        <xs:keyref name="PassengerAssociationRef" refer="tns:InfantIdKey">
                                    <xs:selector xpath="tns:Passenger"/>
                                    <xs:field xpath="@AssociatedInfantID"/>
                        </xs:keyref>
                        <xs:keyref name="InfantAssociationRef" refer="tns:PassengerIdKey">
                                    <xs:selector xpath="tns:Infant"/>
                                    <xs:field xpath="@AssociatedPassengerID"/>
                        </xs:keyref>
            </xs:element>
            <xs:complexType name="PassengerType">
                        <xs:sequence>
                                    <xs:element name="FullName" type="xs:string"/>
                        </xs:sequence>
                        <xs:attribute name="SequenceNo" type="xs:int" use="required"/>
                        <xs:attribute name="ID" type="xs:int" use="required"/>
                        <xs:attribute name="AssociatedInfantID" type="xs:int"/>
            </xs:complexType>
            <xs:complexType name="InfantType">
                        <xs:sequence>
                                    <xs:element name="FullName" type="xs:string"/>
                        </xs:sequence>
                        <xs:attribute name="SequenceNo" type="xs:int" use="required"/>
                        <xs:attribute name="ID" type="xs:int" use="required"/>
                        <xs:attribute name="AssociatedPassengerID" type="xs:int" use="required"/>
            </xs:complexType>                       
</xs:schema>

 

In this schema a list of passengers (Passengers) is defined and this list consists of Passenger elements and Infant elements. Each item(either Passenger or Infant) in the list has a SequenceNo as an attribute. This SequenceNo needs to be unique within the list.  From the schema this can be achieved by defining xs:unique within Passengers definition.

<xs:unique name="PassengerInfantUniqueKey">
      <xs:selector xpath="tns:Passenger | tns:Infant"/>
     <xs:field xpath="@SequenceNo"/>
</xs:unique>

This definition says the SequenceNo attribute in Passenger and Infant needs to be unique within Passengers list where this unique is defined. Therefore the following XML document is not valid because Infant with ID 2 has the duplicate SequenceNo as one in Passenger with ID 1.

<tns:Passengers xmlns:tns="http://www.toic.com/cdm"  instance">
      <tns:Passenger AssociatedInfantID="3" ID="1" SequenceNo="1">
            <tns:FullName>Mark Sean</tns:FullName>
      </tns:Passenger>
      <tns:Passenger ID="2" SequenceNo="2">
            <tns:FullName>John Smith</tns:FullName>
      </tns:Passenger>
      <tns:Infant ID="3" AssociatedPassengerID="1" SequenceNo="1">
            <tns:FullName>Daniel Kemp</tns:FullName>
      </tns:Infant>
</tns:Passengers>

In order to fix this issue we just need to change the value of @SequenceNo in either of two Passengers to a unique value.

 

If you want to keep the association consistency in the list xs:key and xs:keyref can be used to force such consistency.  It is like foreign key referential integrity in RDBMS.   In this example in the list of Passengers one Passenger element may be associated with one Infant element and vice versa.  Such association consistency can achieve by introducing xs:key and xs:keyref elements in Passengers.   In order to do so the first step is to define the Keys and then the KeyRef which will refer the Keys defined before.

As shown below two Keys are defined: one is PassengerIdKey which use the attribute ID in Passenger element as the key and another is InfantIdKey.

<xs:key name="PassengerIdKey">
      <xs:selector xpath="tns:Passenger"/>
      <xs:field xpath="@ID"/>
</xs:key>
 
<xs:key name="InfantIdKey">
      <xs:selector xpath="tns:Infant"/>
      <xs:field xpath="@ID"/>
</xs:key>

Similar to the primary key definition in RDBMS the above definition says we select the attribute @ID in Passenger element as the key: PassengerIdKey and the attribute @ID in Infant element as the key: InfantIdKey.   Of course the key can be also defined base the element field or even a composite key can be defined based on the multiple fields of the element by using more than one xs:field element.

After the keys are defined we can refer these keys by defining keyref.  In the below example keyref PassengerAssociationRef and InfantAssociationRef.   Let us explain this definition by using example PassengerAssociationRef.   It says that the attribute @AssociatedInfantID in element Passenger is used as the keyref and it refers to the key InfantIdKey in Infant.  So the value of attribute AssociatedInfantID should be the same as the value of attribute ID in the associated Infant element.   Otherwise it is not valid against the schema.

<xs:keyref name="PassengerAssociationRef" refer="tns:InfantIdKey">
      <xs:selector xpath="tns:Passenger"/>
      <xs:field xpath="@AssociatedInfantID"/>
</xs:keyref>
 
<xs:keyref name="InfantAssociationRef" refer="tns:PassengerIdKey">
      <xs:selector xpath="tns:Infant"/>
      <xs:field xpath="@AssociatedPassengerID"/>
</xs:keyref>

 

The below is one example of XML document which is not valid against the schema because of the error in the association inconsistency.   The Passenger with ID=1 has the association with one Infant.   From the document this Passenger is supposed to associate with Infant with ID=2.   But actually there is no Infant with ID=2.   If we change AssociateInfantID=”2” in Passenger with ID=”1” to AssociateInfantID=”3” the document becomes valid against the schema.

<tns:Passengers xmlns:tns="http://www.toic.com/cdm">
      <tns:Passenger AssociatedInfantID="2" ID="1" SequenceNo="1">
            <tns:FullName>Mark Sean</tns:FullName>
      </tns:Passenger>
      <tns:Passenger ID="2" SequenceNo="2">
            <tns:FullName>John Smith</tns:FullName>
      </tns:Passenger>
      <tns:Infant ID="3" AssociatedPassengerID="1" SequenceNo="3">
            <tns:FullName>Daniel Kemp</tns:FullName>
      </tns:Infant>
</tns:Passengers>

 

One thing is needed to note that xs:unique, xs:key and xs:ketref are used in element definition rather than type definition.   And also the uniqueness and association consistency forced by the xs:unique, xs:key and xs:keyref are only effective with the element where these are defined.  Here a modified schema will demonstrate this.

<?xml version="1.0" encoding="UTF-8"?>
<xs:schema xmlns:xs="http://www.w3.org/2001/XMLSchema" xmlns:tns="http://www.toic.com/cdm" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" targetNamespace="http://www.toic.com/cdm" elementFormDefault="qualified" attributeFormDefault="unqualified">
                <xs:element name="Flight">
                                <xs:complexType>
                                                <xs:sequence>
                                                                <xs:element ref="tns:Passengers" maxOccurs="unbounded"/>
                                                </xs:sequence>
                                </xs:complexType>
                                <xs:unique name="PassengerInfantUniqueSequnceNo">
                                                <xs:selector xpath="tns:Passengers/tns:Passenger | tns:Passengers/tns:Infant"/>
                                                <xs:field xpath="@SequenceNo"/>
                                </xs:unique>                    
                </xs:element>
               
                <xs:complexType name="PassengerListType">
                                <xs:choice maxOccurs="unbounded">
                                                                <xs:element name="Passenger" type="tns:PassengerType"/>
                                                                <xs:element name="Infant" type="tns:InfantType"/>
                                </xs:choice>
                                <xs:attribute name="CabinClass" type="xs:string" use="required"/>
                </xs:complexType>
               
                <xs:element name="Passengers" type="tns:PassengerListType">
                                <xs:unique name="PassengerInfantUniqueID">
                                                <xs:selector xpath="tns:Passenger | tns:Infant"/>
                                                <xs:field xpath="@ID"/>
                                </xs:unique>                    
                                <xs:key name="PassengerIdKey">
                                                <xs:selector xpath="tns:Passenger"/>
                                                <xs:field xpath="@ID"/>
                                </xs:key>
                                <xs:key name="InfantIdKey">
                                                <xs:selector xpath="tns:Infant"/>
                                                <xs:field xpath="@ID"/>
                                </xs:key>
                                <xs:keyref name="PassengerAssociationRef" refer="tns:InfantIdKey">
                                                <xs:selector xpath="tns:Passenger"/>
                                                <xs:field xpath="@AssociatedInfantID"/>
                                </xs:keyref>
                                <xs:keyref name="InfantAssociationRef" refer="tns:PassengerIdKey">
                                                <xs:selector xpath="tns:Infant"/>
                                                <xs:field xpath="@AssociatedPassengerID"/>
                                </xs:keyref>
                </xs:element>
                <xs:complexType name="PassengerType">
                                <xs:sequence>
                                                <xs:element name="FullName" type="xs:string"/>
                                </xs:sequence>
                                <xs:attribute name="SequenceNo" type="xs:int" use="required"/>
                                <xs:attribute name="ID" type="xs:int" use="required"/>
                                <xs:attribute name="AssociatedInfantID" type="xs:int"/>
                </xs:complexType>
                <xs:complexType name="InfantType">
                                <xs:sequence>
                                                <xs:element name="FullName" type="xs:string"/>
                                </xs:sequence>
                                <xs:attribute name="SequenceNo" type="xs:int" use="required"/>
                                <xs:attribute name="ID" type="xs:int" use="required"/>
                                <xs:attribute name="AssociatedPassengerID" type="xs:int" use="required"/>
                </xs:complexType>       
                <xs:element name="Infant" type="tns:InfantType">
                                <xs:unique name="InfantIDUnique">
                                                <xs:selector xpath="."></xs:selector>
                                                <xs:field xpath="@ID"></xs:field>
                                </xs:unique>
                </xs:element>
</xs:schema>

 

In this new schema there is new element called: Flight.   Each Flight may have one or more than one list of Passengers.   Now the unique PassengerInfantUniqueSequnceNo is moved to element Flight.   It means that values of @SequenceNo for Passenger or Infant need to be unique within Flight no matter what Passengers list it belongs to.   However the value of @ID just needs to be unique with each Passengers list. 

The below XML document is valid even though some Passengers have the duplicate IDs within the Flight.

<?xml version="1.0" encoding="UTF-8"?>
<!--Sample XML file generated by XMLSpy v2008 sp1 (http://www.altova.com)-->
<tns:Flight xsi:schemaLocation="http://www.toic.com/cdm Untitled18.xsd" xmlns:tns="http://www.toic.com/cdm" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance">
                <tns:Passengers CabinClass="A">
                                <tns:Passenger AssociatedInfantID="3" ID="1" SequenceNo="1">
                                                <tns:FullName>String</tns:FullName>
                                </tns:Passenger>
                                <tns:Passenger ID="2" SequenceNo="2">
                                                <tns:FullName>String</tns:FullName>
                                </tns:Passenger>
                               
                                <tns:Infant ID="3" AssociatedPassengerID="1" SequenceNo="3">
                                                <tns:FullName>String</tns:FullName>
                                </tns:Infant>
                </tns:Passengers>
               
                <tns:Passengers CabinClass="B">
                                <tns:Passenger ID="1" SequenceNo="4">
                                                <tns:FullName>String</tns:FullName>
                                </tns:Passenger>
                                <tns:Passenger ID="2" AssociatedInfantID="3"  SequenceNo="5">
                                                <tns:FullName>String</tns:FullName>
                                </tns:Passenger>
                                <tns:Infant ID="3" AssociatedPassengerID="2" SequenceNo="6">
                                                <tns:FullName>String</tns:FullName>
                                </tns:Infant>                   
                </tns:Passengers>
</tns:Flight>

 

 

No comments:

Post a Comment