Class DomXPathRule

  • All Implemented Interfaces:
    Rule, PropertySource

    public class DomXPathRule
    extends AbstractRule
    XPath rule that executes an expression on the DOM directly, and not on the PMD AST wrapper. The XPath expressions adheres to the XPath (2.0) spec, so they can be tested in any existing XPath testing tools instead of just the PMD designer (google "xpath test"). Usage of this class is strongly recommended over the standard XPathRule, which is mostly useful in other PMD languages.

    Differences with XPathRule

    This rule and XPathRule do not accept exactly the same queries, because XPathRule implements the XPath spec in an ad-hoc way. The main differences are:
    • XPathRule uses elements to represent text nodes. This is contrary to the XPath spec, in which element and text nodes are different kinds of nodes. To replace the query //elt/text[@Text="abc"], use the XPath function text(), eg //elt[text()="abc"].
    • XPathRule adds additional attributes to each element (eg @BeginLine and @Text). These attributes are not XML attributes, so they are not accessible using DomXPathRule rule. Instead, use the XPath functions pmd:startLine(node), pmd:endLine(node) and related. For instance, replace //elt[@EndLine - @BeginLine > 10] with elt[pmd:endLine(.) - pmd:startLine(.) > 10].
    • XPathRule uses an element called "document" as the root node of every XML AST. This node does not have the correct node kind, as it's an element, not a document. To replace /document/RootNode, use just /RootNode.
    • XPathRule ignores comments and processing instructions (eg FXML's <?import javafx.Node ?>). This rule makes them accessible with the regular XPath syntax. The following finds all comments in the file:
      
        //comment()
       
      The following finds only top-level comments starting with "prefix":
      
        /comment()[fn:starts-with(fn:string(.), "prefix")]
       
      Note the use of fn:string. As an example of matching processing instructions, the following fetches all <?import ... ?> processing instructions.
      
        /processing-instruction('import')
       
      The string value of the instruction can be found with fn:string.

    Additionally, this rule only supports XPath 2.0, with no option for configuration. This will be bumped to XPath 3.1 in PMD 7.

    Namespace-sensitivity

    Another important difference is that this rule is namespace-sensitive. If the tested XML documents use a schema (xmlns attribute on the root), you should set the property defaultNsUri on the rule with the value of the xmlns attribute. Otherwise node tests won't match unless you use a wildcard URI prefix (*:nodeName).

    For instance for the document

    
     <foo xmlns="http://company.com/aschema">
     </foo>
     
    the XPath query //foo will not match anything, while //*:foo will. If you set the property defaultNsUri to "http://company.com/aschema", then //foo will be expanded to //Q{http://company.com/aschema}foo, and match the foo node. The behaviour is equivalent in the following document:
    
     <my:foo xmlns:my='http://company.com/aschema'>
     </my:foo>
     

    However, for the document

    
     <foo>
     </foo>
     
    the XPath queries //foo and //*:foo both match, because //foo is expanded to //Q{}foo (local name foo, empty URI), and the document has no default namespace (= the empty default namespace).

    Note that explicitly specifying URIs with Q{...}localName as in this documentation is XPath 3.1 syntax and will only be available in PMD 7.

    Author:
    Clément Fournier
    Since:
    PMD 6.44.0
    • Constructor Detail

      • DomXPathRule

        public DomXPathRule()
      • DomXPathRule

        public DomXPathRule​(String xpath)
      • DomXPathRule

        public DomXPathRule​(String xpath,
                            String defaultNsUri)