Querying HTML and XML Documents

Zend\Dom\Query provides mechanisms for querying XML and HTML documents utilizing either XPath or CSS selectors. It was developed to aid with functional testing of MVC applications, but could also be used for development of screen scrapers.

CSS selector notation is provided as a simpler and more familiar notation for web developers to utilize when querying documents with XML structures. The notation should be familiar to anybody who has developed Cascading Style Sheets or who utilizes javascript toolkits that provide functionality for selecting nodes utilizing CSS selectors. Prototype's $$(), Dojo's dojo.query, and jQuery were all inspirations for the component.

Theory of Operation

To use Zend\Dom\Query, you instantiate a Zend\Dom\Query object, optionally passing a document to query (a string). Once you have a document, you can use either the execute() or queryXpath() methods; each method will return a Zend\Dom\NodeList object with any matching nodes.

The primary difference between Zend\Dom\Query and using DOMDocument + DOMXPath is the ability to select against CSS + selectors. You can utilize any of the following, in any combination:

• element types: provide an element type to match: div, a, span, h2, etc.
• style attributes: CSS style attributes to match: .error, div.error, label.required, etc. If an element defines more than one style, this will match as long as the named style is present anywhere in the style declaration.
• id attributes: element ID attributes to match: #content, div#nav, etc.
• arbitrary attributes: arbitrary element attributes to match. Three different types of matching are provided:
  • exact match: the attribute exactly matches the specified string. div[bar="baz"] would match a div element with a bar attribute that exactly matches the value baz.
  • word match: the attribute contains a word matching the string. div[bar~="baz"] would match a div element with a bar attribute that contains the word baz. <div bar="foo baz"> would match, but <div bar="foo bazbat"> would not.
  • substring match: the attribute contains the string specified, whether or not it is a complete word. div[bar*="baz"] would match a div element with a bar attribute that contains the string baz anywhere within it.
• direct descendents: utilize > between selectors to denote direct descendents. div > span would select only span elements that are direct descendents of a div. Can also be used with any of the selectors above.
• descendents: string together multiple selectors to indicate a hierarchy along which to search. div .foo span #one would select an element of id one that is a descendent of arbitrary depth beneath a span element, which is in turn a descendent of arbitrary depth beneath an element with a class of foo, that is an descendent of arbitrary depth beneath a div element. For example, it would match the link to the word 'One' in the listing below:

    <div>
    <table>
        <tr>
            <td class="foo">
                <div>
                    Lorem ipsum <span class="bar">
                        <a href="/foo/bar" id="one">One</a>
                        <a href="/foo/baz" id="two">Two</a>
                        <a href="/foo/bat" id="three">Three</a>
                        <a href="/foo/bla" id="four">Four</a>
                    </span>
                </div>
            </td>
        </tr>
    </table>
    </div>

Once you've performed your query, you can then work with the result object to determine information about the nodes, as well as to pull them and/or their content directly for examination and manipulation. Zend\Dom\NodeList implements Countable and Iterator, and stores the results internally as a DOMDocument and DOMNodeList.

As an example, consider the following call, that selects against the HTML above:

use Zend\Dom\Query;

$dom = new Query($html);
$results = $dom->execute('.foo .bar a');

$count = count($results); // get number of matches: 4
foreach ($results as $result) {
    // $result is a DOMElement
}

Zend\Dom\Query also allows straight XPath queries utilizing the queryXpath() method; you can pass any valid XPath query to this method, and it will return a Zend\Dom\NodeList object.

Methods Available

Below is a listing of methods available in the various classes exposed by zend-dom.

Zend\Dom\Query

The following methods are available to Zend\Dom\Query:

• setDocumentXml($document, $encoding = null): specify an XML string to query against.
• setDocumentXhtml($document, $encoding = null): specify an XHTML string to query against.
• setDocumentHtml($document, $encoding = null): specify an HTML string to query against.
• setDocument($document, $encoding = null): specify a string to query against; Zend\Dom\Query will then attempt to autodetect the document type.
• setEncoding($encoding): specify an encoding string to use. This encoding will be passed to DOMDocument's constructor if specified.
• getDocument(): retrieve the original document string provided to the object.
• getDocumentType(): retrieve the document type of the document provided to the object; will be one of the DOC_XML, DOC_XHTML, or DOC_HTML class constants.
• getEncoding(): retrieves the specified encoding.
• execute($query): query the document using CSS selector notation.
• queryXpath($xPathQuery): query the document using XPath notation.

Zend\Dom\NodeList

As mentioned previously, Zend\Dom\NodeList implements both Iterator and Countable, and as such can be used in a foreach() loop as well as with the count() function. Additionally, it exposes the following methods:

• getCssQuery(): return the CSS selector query used to produce the result (if any).
• getXpathQuery(): return the XPath query used to produce the result. Internally, Zend\Dom\Query converts CSS selector queries to XPath, so this value will always be populated.
• getDocument(): retrieve the DOMDocument the selection was made against.