XPath Requirements
Version 2.0

The following examples illustrate limitations of XPath 1.0 that can be addressed by the addition of existential and universal quantifiers to XPath 2.0:

1. In XPath 1.0, the expression 5 > $ns evaluates to true if 5 is greater than the number() value of at least one node in the $ns node set. There is no way to test whether 5 is greater than the number() values of all nodes in $ns.
2. In XPath 1.0, the expression $ns1 > $ns2 evaluates to true if the number() value of at least one node in $ns1 is greater than the number() value of at least one node in the $ns2 node set. There is no way to test whether the number() values of all of the nodes in $ns1 are greater than all the number() values of all nodes in $ns2.
3. In XPath 1.0, the expression 5 = $ns evaluates to true if 5 equals the number() value of at least one node in the $ns node set. There is no way to test whether 5 equals the number() value of all nodes in $ns.
4. In XPath 1.0, the expression $ns1 = $ns2 evaluates to true if the number() value of at least one node in $ns1 equals the number() value of at least one node in the $ns2 node set. There is no way to test whether the number() values of all of the nodes in $ns1 are greater than all the number() values of all nodes in $ns2.
5. In XPath 1.0, the expression 0 + LineItem/@UnitPrice > 40 is equivalent to 0 + number(OrderDetail/@UnitPrice) > 40 which will be true if the first LineItem element's UnitPrice attribute is greater than 40. However, the expression LineItem/@UnitPrice > 40 is true if any of the LineItem elements' UnitPrice attributes is greater than 40. This difference in behavior could be resolved through explicit quantification.

1. Given an XML document like:

   <hr>
      <teams>
       <team id="t1">
         <members ref="e101 e103"/>
       </team>
       <team id="t2">
         <members ref="e102"/>
       </team>
     </teams>
     <departments>
       <department>
         <employee id="e101" name="Steve" teams="t1"/>
         <employee id="e102" name="Jonathan" teams="t2"/>
         <employee id="e103" name="James" teams="t1"/>
       </department>
     </departments>
   </hr>
   
   

   Format the list of employees that work in the same department as the employee with id = e101 and that are on the same teams as that employee. To accomplish this:

   • Select the list of all employees in the employee's department:

     id('e101')/parent::department/employee
     
     

   • Intersect that set with the list of all employees who are members of the current employee's teams:

     id('e101')/id(id(@teams)/members/@ref)
     
     

   • Difference that set with the list containing the current employee so they don't appear in their own list of team members:

     id('e101')
     
     

   The result should be the employee "James".

2. Given an XML document full of book information, a book with multiple authors appears like this:

   <books>
     <book>
       <title>Exciting New Book</title>
       <author>Kay</author>
       <author>Muench</author>
     </book>
     <book>
       <title>Building Oracle XML Applications</title>
       <author>Muench</author>
     </book>
     <book>
       <title>XSLT Programmer's Reference</title>
       <author>Kay</author>
     </book>
   </books>
   
   

   Assuming an XSLT key is defined as follows:

   <xsl:key name="auth" match="book" use="author"/>
   
   

   then, to find the list of books that have both Muench and Kay among their authors, you need to select the intersection of key('auth','Muench') and key('auth','Kay').

   To find books authored by Muench but not by Kay, you need to select the difference of key('auth','Muench') and key('auth','Kay').

1. An XPointer's expression can use node-set functions on the right of the '/' in a path:

   id("chap1")/range-to(id("chap2"))
   
   

2. Allowing unions in a location step would allow expressions like:

   /foo/(xxx|yyy)/li/(p|eg)
   
   

   instead of the more cumbersome:

   /foo/xxx/li/p  |
   /foo/xxx/li/eg |
   /foo/yyy/li/p  |
   /foo/yyy/li/eg
   
   

3. Allowing node-set functions to appear in a location step would allow expressions like:

   document("otherdoc.xml")/id("foo")/a/b
   
   

   and

   document("otherdoc.xml")/key("x","foo")/a/b
   
   

XSLT Use Cases

1. In a template, construct a <table> element with a border attribute whose value is provided by an attribute value template and is conditional depending on the value of the frame attribute of the current node. If @frame = 'none', then the value of border should be 0, otherwise the value of border should be 1.
2. Assign an XSLT variable a conditional value based on whether the current node has any <test> element children. If it does, assign the value "is" to the variable, otherwise assign the value "is not".

3. Given a source XML document like this:

   <Emps>
     <Emp>
        <LastName>Smith</LastName>
     </Emp>
     <Emp>
       <CommonName>Scott</CommonName>
     </Emp>
   </Emps>
   
   

   provide an <xsl:sort> select expression to sort the list of employees on the value of their <LastName> if it's present, otherwise on the value of their <CommonName> element.

1. In XPath 1.0, the expression a[b=5] returns <a> elements that have ANY <b> element child with value 5, where as the expression a[b+1=6] returns <a> elements whose FIRST <b> element child has value 5. These results should be consistent for XPath 2.0.

   In 2.0 a[b+1=6] will return an error if more than one b where in 1.0 it will ignore the other b's. This produces a less surprising reqsult for our users.

2. Assume for this use case that the typed-value of an element or attribute is referenced using the syntax typed-value(elt) and typed-value(@attr), respectively. In XPath 2.0, when working with the typed value of an element or attribute with boolean type, the expression a[typed-value(b) = false()] should return <a> elements having at least one <b> element child with boolean value false. Similarly, the expression a[typed-value(@b) = false()] should return <a> elements having a boolean attribute b with value false.

Replace each occurrence of "foo" in a string with "foobar".

1. Pad the value of a string on the right to ensure a length of 10 characters using spaces
2. Pad the value of a string on the left to 10 characters using asterisks
3. Output a string containing as many dashes as the length of a section title

1. Given a document like

   <sect title="Plan">
      <sect title="Overview">
         <sect title="Competitors">
           <sect title="SuperSoft"/>
           <sect title="BarnWare"/>
         </sect>
      </sect>
      <sect title="Details">
         <sect title="Summary"/>
      </sect>
   </sect>
   
   

   Calculate the maximum depth of the nesting of <sect> elements. The nodeset is identified by the expression //sect and the expression to maximize would be count(ancestor-or-self::sect).

2. Given the same document above, calculate the average string length of section titles. The nodeset is identified by //section/@title and the expression to sum would be string-length(.), divided by the count(//section/@title)
3. Given an XML document containing a purchase order and its line <item> elements, calculate the total amount of the purchase order by summing the price times the quantity of each item. The nodeset is identified by item, and the expression to sum would be price * quantity.

1. Given XML like:

   <Data>
     <EmpRow>
        :
     </EmpRow>
     <DeptRow>
        <Deptno>10</Deptno>
        <Employees>
           <EmpRow>
             <Empno>1000</Empno>
             <Ename>1000</Ename>
           </Emprow>
        </Employees>
     </DeptRow>
   </Data>
   
   

   match elements that "end with Row or row"

   *[local-name() =~ ".*[rR]ow"]
   
   

2. Match phone numbers like 123-456-7890

   field[ . =~ "\d\d\d-\d\d\d\-\d\d\d\d]
   
   

3. Match any element name with the string "Addr" contained within the name (case-insensitively)

   *[local-name() =~ ".*[Aa][Dd][Dd][Rr].*"]
   
   

4. Match a string from a dynamic regular expression

   Department[Code =~ concat("[0-9][0-9]",$v,".*")]
   
   

• General

  1. Use and create the type information of any datatype
  2. Do not require the user to respecify the datatype (i.e. number and boolean)
  3. Maintain the facet constraints of primitive datatypes while manipulating the value. For example, take a 14.4 number plus a 14.4 number and return a 14.4 number.

• For DateTime

  1. Compare dates of documents and return a boolean
  2. Add a time or date interval and return the same type
  3. Add a scalar value to a time or date and return the same type

  4. Calculate an end time from a start time+duration. For example, given:

     <appointment time="20001224-16:00:00" duration="1:00"/>

     Produce the end-time:

     End time: <xsl:value-of select="@time + @duration"/>
     
     

  5. Detect intersections between datetime durations

     <appointment time="20001224-16:00:00" duration="2:00" reminder="0:15"/>
     
     
     <appointment time="20001224-17:00:00" duration="1:00" reminder="0:15"/>
     
     

• For Boolean

  Perform pattern matching on "yes" and "no"

• For QName

  1. Build a Namespace element from a QName
  2. Compare a Namespace URI from a QName with a string

Be able to compare an element or attribute with an XML Schema date or time type with a constant date or time value.

Create an XSLT transformation that converts an element structure like this:

into an alternative structure like this:

This requires the ability in XPath to select the list of NMTOKENS values of the dt:values attribute for processing.

Assuming the typed value of an element is addressed using a syntax like typed-value(elementname), XPath needs to specify what 1.20 * typed-value(salary) evaluates to when the typed value of the <salary> is null.

1. Select all elements that are instances of the complex type Address.
2. Select all attributes that are instances of the simple-type xsd:integer.
3. Select elements of type Address indicated in the instance with xsi:type="Address"

Derivation by Restriction

The [XML Schema: Primer] defines the Items type and the ConfirmedItems type which is derived by restriction from Items. Select purchase orders whose <items> member element is an instance of the ConfirmedItems type, which restricts the minimum number of <item> elements that can appear to be at least one.

Derivation by Extension

The [XML Schema: Primer] defines the Address type and the UKAddress and USAddress types which is derived by extension from Address, adding additional element content in the subtypes.

1. Select all elements that are instances of complex type Address, matching elements of type USAddress and elements of type UKAddress as well.
2. Select all elements that are instances of complex type Address, without selecting elements of types derived from Address.
3. Select all attributes that are instances of the simple-type xsd:integer, matching attribute of type SKU-Number which derives from xsd:integer.

The [XML Schema: Primer] defines the global <ipo:comment> element as the head of a substitution group, whose substitutible member elements include <ipo:customerComment> and <ipo:shipComment>. Select the list of all elements that are <ipo:comment> elements, including any elements that are substitutible for <ipo:comment> through this substitution group.